ladybird/Userland/Libraries/LibJS/Runtime/Object.cpp

1411 lines
50 KiB
C++
Raw Normal View History

/*
* Copyright (c) 2020, Andreas Kling <kling@serenityos.org>
* Copyright (c) 2020-2023, Linus Groh <linusg@serenityos.org>
*
* SPDX-License-Identifier: BSD-2-Clause
*/
#include <AK/DeprecatedString.h>
#include <AK/TypeCasts.h>
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
#include <LibJS/Runtime/AbstractOperations.h>
#include <LibJS/Runtime/Accessor.h>
#include <LibJS/Runtime/Array.h>
#include <LibJS/Runtime/ClassFieldDefinition.h>
#include <LibJS/Runtime/ECMAScriptFunctionObject.h>
#include <LibJS/Runtime/Error.h>
#include <LibJS/Runtime/GlobalObject.h>
#include <LibJS/Runtime/NativeFunction.h>
#include <LibJS/Runtime/Object.h>
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
#include <LibJS/Runtime/PropertyDescriptor.h>
#include <LibJS/Runtime/ProxyObject.h>
#include <LibJS/Runtime/Shape.h>
#include <LibJS/Runtime/Value.h>
namespace JS {
static HashMap<GCPtr<Object const>, HashMap<DeprecatedFlyString, Object::IntrinsicAccessor>> s_intrinsics;
// 10.1.12 OrdinaryObjectCreate ( proto [ , additionalInternalSlotsList ] ), https://tc39.es/ecma262/#sec-ordinaryobjectcreate
NonnullGCPtr<Object> Object::create(Realm& realm, Object* prototype)
{
if (!prototype)
return realm.heap().allocate<Object>(realm, realm.intrinsics().empty_object_shape());
if (prototype == realm.intrinsics().object_prototype())
return realm.heap().allocate<Object>(realm, realm.intrinsics().new_object_shape());
return realm.heap().allocate<Object>(realm, ConstructWithPrototypeTag::Tag, *prototype);
}
Object::Object(GlobalObjectTag, Realm& realm)
{
// This is the global object
m_shape = heap().allocate_without_realm<Shape>(realm);
}
Object::Object(ConstructWithoutPrototypeTag, Realm& realm)
{
m_shape = heap().allocate_without_realm<Shape>(realm);
}
Object::Object(Realm& realm, Object* prototype)
{
m_shape = realm.intrinsics().empty_object_shape();
VERIFY(m_shape);
if (prototype != nullptr)
set_prototype(prototype);
}
Object::Object(ConstructWithPrototypeTag, Object& prototype)
{
m_shape = prototype.shape().realm().intrinsics().empty_object_shape();
VERIFY(m_shape);
set_prototype(&prototype);
}
Object::Object(Shape& shape)
: m_shape(&shape)
{
m_storage.resize(shape.property_count());
}
Object::~Object()
{
if (m_has_intrinsic_accessors)
s_intrinsics.remove(this);
}
void Object::initialize(Realm&)
{
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 7.2 Testing and Comparison Operations, https://tc39.es/ecma262/#sec-testing-and-comparison-operations
// 7.2.5 IsExtensible ( O ), https://tc39.es/ecma262/#sec-isextensible-o
ThrowCompletionOr<bool> Object::is_extensible() const
{
// 1. Return ? O.[[IsExtensible]]().
return internal_is_extensible();
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 7.3 Operations on Objects, https://tc39.es/ecma262/#sec-operations-on-objects
// 7.3.2 Get ( O, P ), https://tc39.es/ecma262/#sec-get-o-p
ThrowCompletionOr<Value> Object::get(PropertyKey const& property_key) const
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Return ? O.[[Get]](P, O).
return TRY(internal_get(property_key, this));
}
// NOTE: 7.3.3 GetV ( V, P ) is implemented as Value::get().
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 7.3.4 Set ( O, P, V, Throw ), https://tc39.es/ecma262/#sec-set-o-p-v-throw
ThrowCompletionOr<void> Object::set(PropertyKey const& property_key, Value value, ShouldThrowExceptions throw_exceptions)
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto& vm = this->vm();
VERIFY(property_key.is_valid());
VERIFY(!value.is_empty());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let success be ? O.[[Set]](P, V, O).
auto success = TRY(internal_set(property_key, value, this));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If success is false and Throw is true, throw a TypeError exception.
if (!success && throw_exceptions == ShouldThrowExceptions::Yes) {
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// FIXME: Improve/contextualize error message
return vm.throw_completion<TypeError>(ErrorType::ObjectSetReturnedFalse);
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. Return unused.
return {};
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 7.3.5 CreateDataProperty ( O, P, V ), https://tc39.es/ecma262/#sec-createdataproperty
ThrowCompletionOr<bool> Object::create_data_property(PropertyKey const& property_key, Value value)
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let newDesc be the PropertyDescriptor { [[Value]]: V, [[Writable]]: true, [[Enumerable]]: true, [[Configurable]]: true }.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto new_descriptor = PropertyDescriptor {
.value = value,
.writable = true,
.enumerable = true,
.configurable = true,
};
// 2. Return ? O.[[DefineOwnProperty]](P, newDesc).
return internal_define_own_property(property_key, new_descriptor);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 7.3.6 CreateMethodProperty ( O, P, V ), https://tc39.es/ecma262/#sec-createmethodproperty
void Object::create_method_property(PropertyKey const& property_key, Value value)
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
VERIFY(!value.is_empty());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Assert: O is an ordinary, extensible object with no non-configurable properties.
// 2. Let newDesc be the PropertyDescriptor { [[Value]]: V, [[Writable]]: true, [[Enumerable]]: false, [[Configurable]]: true }.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto new_descriptor = PropertyDescriptor {
.value = value,
.writable = true,
.enumerable = false,
.configurable = true,
};
// 3. Perform ! O.[[DefineOwnProperty]](P, newDesc).
MUST(internal_define_own_property(property_key, new_descriptor));
// 4. Return unused.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 7.3.7 CreateDataPropertyOrThrow ( O, P, V ), https://tc39.es/ecma262/#sec-createdatapropertyorthrow
ThrowCompletionOr<bool> Object::create_data_property_or_throw(PropertyKey const& property_key, Value value)
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
auto& vm = this->vm();
VERIFY(property_key.is_valid());
VERIFY(!value.is_empty());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let success be ? CreateDataProperty(O, P, V).
auto success = TRY(create_data_property(property_key, value));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If success is false, throw a TypeError exception.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!success) {
// FIXME: Improve/contextualize error message
return vm.throw_completion<TypeError>(ErrorType::ObjectDefineOwnPropertyReturnedFalse);
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. Return success.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return success;
}
// 7.3.8 CreateNonEnumerableDataPropertyOrThrow ( O, P, V ), https://tc39.es/ecma262/#sec-createnonenumerabledatapropertyorthrow
void Object::create_non_enumerable_data_property_or_throw(PropertyKey const& property_key, Value value)
{
VERIFY(property_key.is_valid());
VERIFY(!value.is_empty());
// 1. Assert: O is an ordinary, extensible object with no non-configurable properties.
// 2. Let newDesc be the PropertyDescriptor { [[Value]]: V, [[Writable]]: true, [[Enumerable]]: false, [[Configurable]]: true }.
auto new_description = PropertyDescriptor { .value = value, .writable = true, .enumerable = false, .configurable = true };
// 3. Perform ! DefinePropertyOrThrow(O, P, newDesc).
MUST(define_property_or_throw(property_key, new_description));
// 4. Return unused.
}
// 7.3.9 DefinePropertyOrThrow ( O, P, desc ), https://tc39.es/ecma262/#sec-definepropertyorthrow
ThrowCompletionOr<void> Object::define_property_or_throw(PropertyKey const& property_key, PropertyDescriptor const& property_descriptor)
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto& vm = this->vm();
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let success be ? O.[[DefineOwnProperty]](P, desc).
auto success = TRY(internal_define_own_property(property_key, property_descriptor));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If success is false, throw a TypeError exception.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!success) {
// FIXME: Improve/contextualize error message
return vm.throw_completion<TypeError>(ErrorType::ObjectDefineOwnPropertyReturnedFalse);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 3. Return unused.
return {};
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 7.3.10 DeletePropertyOrThrow ( O, P ), https://tc39.es/ecma262/#sec-deletepropertyorthrow
ThrowCompletionOr<void> Object::delete_property_or_throw(PropertyKey const& property_key)
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
auto& vm = this->vm();
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let success be ? O.[[Delete]](P).
auto success = TRY(internal_delete(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If success is false, throw a TypeError exception.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!success) {
// FIXME: Improve/contextualize error message
return vm.throw_completion<TypeError>(ErrorType::ObjectDeleteReturnedFalse);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 3. Return unused.
return {};
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 7.3.12 HasProperty ( O, P ), https://tc39.es/ecma262/#sec-hasproperty
ThrowCompletionOr<bool> Object::has_property(PropertyKey const& property_key) const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Return ? O.[[HasProperty]](P).
return internal_has_property(property_key);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 7.3.13 HasOwnProperty ( O, P ), https://tc39.es/ecma262/#sec-hasownproperty
ThrowCompletionOr<bool> Object::has_own_property(PropertyKey const& property_key) const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let desc be ? O.[[GetOwnProperty]](P).
auto descriptor = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If desc is undefined, return false.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!descriptor.has_value())
return false;
// 3. Return true.
return true;
}
// 7.3.16 SetIntegrityLevel ( O, level ), https://tc39.es/ecma262/#sec-setintegritylevel
ThrowCompletionOr<bool> Object::set_integrity_level(IntegrityLevel level)
{
auto& vm = this->vm();
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let status be ? O.[[PreventExtensions]]().
auto status = TRY(internal_prevent_extensions());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If status is false, return false.
if (!status)
return false;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. Let keys be ? O.[[OwnPropertyKeys]]().
auto keys = TRY(internal_own_property_keys());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 4. If level is sealed, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (level == IntegrityLevel::Sealed) {
// a. For each element k of keys, do
for (auto& key : keys) {
auto property_key = MUST(PropertyKey::from_value(vm, key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// i. Perform ? DefinePropertyOrThrow(O, k, PropertyDescriptor { [[Configurable]]: false }).
TRY(define_property_or_throw(property_key, { .configurable = false }));
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 5. Else,
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
else {
// a. Assert: level is frozen.
// b. For each element k of keys, do
for (auto& key : keys) {
auto property_key = MUST(PropertyKey::from_value(vm, key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// i. Let currentDesc be ? O.[[GetOwnProperty]](k).
auto current_descriptor = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// ii. If currentDesc is not undefined, then
if (!current_descriptor.has_value())
continue;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
PropertyDescriptor descriptor;
// 1. If IsAccessorDescriptor(currentDesc) is true, then
if (current_descriptor->is_accessor_descriptor()) {
// a. Let desc be the PropertyDescriptor { [[Configurable]]: false }.
descriptor = { .configurable = false };
}
// 2. Else,
else {
// a. Let desc be the PropertyDescriptor { [[Configurable]]: false, [[Writable]]: false }.
descriptor = { .writable = false, .configurable = false };
}
// 3. Perform ? DefinePropertyOrThrow(O, k, desc).
TRY(define_property_or_throw(property_key, descriptor));
}
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 6. Return true.
return true;
}
// 7.3.17 TestIntegrityLevel ( O, level ), https://tc39.es/ecma262/#sec-testintegritylevel
ThrowCompletionOr<bool> Object::test_integrity_level(IntegrityLevel level) const
{
auto& vm = this->vm();
// 1. Let extensible be ? IsExtensible(O).
auto extensible = TRY(is_extensible());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If extensible is true, return false.
// 3. NOTE: If the object is extensible, none of its properties are examined.
if (extensible)
return false;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 4. Let keys be ? O.[[OwnPropertyKeys]]().
auto keys = TRY(internal_own_property_keys());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 5. For each element k of keys, do
for (auto& key : keys) {
auto property_key = MUST(PropertyKey::from_value(vm, key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// a. Let currentDesc be ? O.[[GetOwnProperty]](k).
auto current_descriptor = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. If currentDesc is not undefined, then
if (!current_descriptor.has_value())
continue;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// i. If currentDesc.[[Configurable]] is true, return false.
if (*current_descriptor->configurable)
return false;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// ii. If level is frozen and IsDataDescriptor(currentDesc) is true, then
if (level == IntegrityLevel::Frozen && current_descriptor->is_data_descriptor()) {
// 1. If currentDesc.[[Writable]] is true, return false.
if (*current_descriptor->writable)
return false;
}
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 6. Return true.
return true;
}
// 7.3.24 EnumerableOwnPropertyNames ( O, kind ), https://tc39.es/ecma262/#sec-enumerableownpropertynames
ThrowCompletionOr<MarkedVector<Value>> Object::enumerable_own_property_names(PropertyKind kind) const
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// NOTE: This has been flattened for readability, so some `else` branches in the
// spec text have been replaced with `continue`s in the loop below.
auto& vm = this->vm();
auto& realm = *vm.current_realm();
// 1. Let ownKeys be ? O.[[OwnPropertyKeys]]().
auto own_keys = TRY(internal_own_property_keys());
// 2. Let properties be a new empty List.
auto properties = MarkedVector<Value> { heap() };
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. For each element key of ownKeys, do
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
for (auto& key : own_keys) {
// a. If Type(key) is String, then
if (!key.is_string())
continue;
auto property_key = MUST(PropertyKey::from_value(vm, key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// i. Let desc be ? O.[[GetOwnProperty]](key).
auto descriptor = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// ii. If desc is not undefined and desc.[[Enumerable]] is true, then
if (descriptor.has_value() && *descriptor->enumerable) {
// 1. If kind is key, append key to properties.
if (kind == PropertyKind::Key) {
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
properties.append(key);
continue;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. Else,
// a. Let value be ? Get(O, key).
auto value = TRY(get(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. If kind is value, append value to properties.
if (kind == PropertyKind::Value) {
properties.append(value);
continue;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// c. Else,
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// i. Assert: kind is key+value.
VERIFY(kind == PropertyKind::KeyAndValue);
// ii. Let entry be CreateArrayFromList(« key, value »).
auto entry = Array::create_from(realm, { key, value });
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// iii. Append entry to properties.
properties.append(entry);
}
}
// 4. Return properties.
return { move(properties) };
}
// 7.3.26 CopyDataProperties ( target, source, excludedItems ), https://tc39.es/ecma262/#sec-copydataproperties
ThrowCompletionOr<void> Object::copy_data_properties(VM& vm, Value source, HashTable<PropertyKey> const& seen_names)
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
{
2023-04-14 17:05:02 +02:00
// 1. If source is either undefined or null, return unused.
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (source.is_nullish())
return {};
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
2023-04-14 17:05:02 +02:00
// 2. Let from be ! ToObject(source).
auto from = MUST(source.to_object(vm));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
2023-04-14 17:05:02 +02:00
// 3. Let keys be ? from.[[OwnPropertyKeys]]().
auto keys = TRY(from->internal_own_property_keys());
// 4. For each element nextKey of keys, do
for (auto& next_key_value : keys) {
auto next_key = MUST(PropertyKey::from_value(vm, next_key_value));
2023-04-14 17:05:02 +02:00
// a. Let excluded be false.
// b. For each element e of excludedItems, do
// i. If SameValue(e, nextKey) is true, then
// 1. Set excluded to true.
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (seen_names.contains(next_key))
continue;
2023-04-14 17:05:02 +02:00
// c. If excluded is false, then
// i. Let desc be ? from.[[GetOwnProperty]](nextKey).
auto desc = TRY(from->internal_get_own_property(next_key));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
2023-04-14 17:05:02 +02:00
// ii. If desc is not undefined and desc.[[Enumerable]] is true, then
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
if (desc.has_value() && desc->attributes().is_enumerable()) {
2023-04-14 17:05:02 +02:00
// 1. Let propValue be ? Get(from, nextKey).
auto prop_value = TRY(from->get(next_key));
// 2. Perform ! CreateDataPropertyOrThrow(target, nextKey, propValue).
MUST(create_data_property_or_throw(next_key, prop_value));
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
}
2023-04-14 17:05:02 +02:00
// 5. Return unused.
return {};
LibJS: Make scoping follow the spec Before this we used an ad-hoc combination of references and 'variables' stored in a hashmap. This worked in most cases but is not spec like. Additionally hoisting, dynamically naming functions and scope analysis was not done properly. This patch fixes all of that by: - Implement BindingInitialization for destructuring assignment. - Implementing a new ScopePusher which tracks the lexical and var scoped declarations. This hoists functions to the top level if no lexical declaration name overlaps. Furthermore we do checking of redeclarations in the ScopePusher now requiring less checks all over the place. - Add methods for parsing the directives and statement lists instead of having that code duplicated in multiple places. This allows declarations to pushed to the appropriate scope more easily. - Remove the non spec way of storing 'variables' in DeclarativeEnvironment and make Reference follow the spec instead of checking both the bindings and 'variables'. - Remove all scoping related things from the Interpreter. And instead use environments as specified by the spec. This also includes fixing that NativeFunctions did not produce a valid FunctionEnvironment which could cause issues with callbacks and eval. All FunctionObjects now have a valid NewFunctionEnvironment implementation. - Remove execute_statements from Interpreter and instead use ASTNode::execute everywhere this simplifies AST.cpp as you no longer need to worry about which method to call. - Make ScopeNodes setup their own environment. This uses four different methods specified by the spec {Block, Function, Eval, Global}DeclarationInstantiation with the annexB extensions. - Implement and use NamedEvaluation where specified. Additionally there are fixes to things exposed by these changes to eval, {for, for-in, for-of} loops and assignment. Finally it also fixes some tests in test-js which where passing before but not now that we have correct behavior :^).
2021-09-22 12:44:56 +02:00
}
// 7.3.27 PrivateElementFind ( O, P ), https://tc39.es/ecma262/#sec-privateelementfind
PrivateElement* Object::private_element_find(PrivateName const& name)
{
if (!m_private_elements)
return nullptr;
2023-04-14 17:05:02 +02:00
// 1. If O.[[PrivateElements]] contains a PrivateElement pe such that pe.[[Key]] is P, then
auto it = m_private_elements->find_if([&](auto const& element) {
return element.key == name;
});
2023-04-14 17:05:02 +02:00
if (!it.is_end()) {
// a. Return pe.
return &(*it);
}
2023-04-14 17:05:02 +02:00
// 2. Return empty.
return nullptr;
}
// 7.3.28 PrivateFieldAdd ( O, P, value ), https://tc39.es/ecma262/#sec-privatefieldadd
ThrowCompletionOr<void> Object::private_field_add(PrivateName const& name, Value value)
{
auto& vm = this->vm();
// 1. If the host is a web browser, then
// a. Perform ? HostEnsureCanAddPrivateElement(O).
// NOTE: Since LibJS has no way of knowing whether it is in a browser we just always call the hook.
TRY(vm.host_ensure_can_add_private_element(*this));
// 2. Let entry be PrivateElementFind(O, P).
// 3. If entry is not empty, throw a TypeError exception.
if (auto* entry = private_element_find(name); entry)
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldAlreadyDeclared, name.description);
if (!m_private_elements)
m_private_elements = make<Vector<PrivateElement>>();
// 4. Append PrivateElement { [[Key]]: P, [[Kind]]: field, [[Value]]: value } to O.[[PrivateElements]].
m_private_elements->empend(name, PrivateElement::Kind::Field, make_handle(value));
// 5. Return unused.
return {};
}
// 7.3.29 PrivateMethodOrAccessorAdd ( O, method ), https://tc39.es/ecma262/#sec-privatemethodoraccessoradd
ThrowCompletionOr<void> Object::private_method_or_accessor_add(PrivateElement element)
{
auto& vm = this->vm();
// 1. Assert: method.[[Kind]] is either method or accessor.
VERIFY(element.kind == PrivateElement::Kind::Method || element.kind == PrivateElement::Kind::Accessor);
// 2. If the host is a web browser, then
// a. Perform ? HostEnsureCanAddPrivateElement(O).
// NOTE: Since LibJS has no way of knowing whether it is in a browser we just always call the hook.
TRY(vm.host_ensure_can_add_private_element(*this));
// 3. Let entry be PrivateElementFind(O, method.[[Key]]).
// 4. If entry is not empty, throw a TypeError exception.
if (auto* entry = private_element_find(element.key); entry)
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldAlreadyDeclared, element.key.description);
if (!m_private_elements)
m_private_elements = make<Vector<PrivateElement>>();
// 5. Append method to O.[[PrivateElements]].
m_private_elements->append(move(element));
// 6. Return unused.
return {};
}
2023-04-14 17:05:02 +02:00
// 7.3.31 PrivateGet ( O, P ), https://tc39.es/ecma262/#sec-privateget
ThrowCompletionOr<Value> Object::private_get(PrivateName const& name)
{
auto& vm = this->vm();
2023-04-14 17:05:02 +02:00
// 1. Let entry be PrivateElementFind(O, P).
auto* entry = private_element_find(name);
2023-04-14 17:05:02 +02:00
// 2. If entry is empty, throw a TypeError exception.
if (!entry)
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldDoesNotExistOnObject, name.description);
auto& value = entry->value;
2023-04-14 17:05:02 +02:00
// 3. If entry.[[Kind]] is either field or method, then
if (entry->kind != PrivateElement::Kind::Accessor) {
// a. Return entry.[[Value]].
return value.value();
2023-04-14 17:05:02 +02:00
}
2023-04-14 17:05:02 +02:00
// Assert: entry.[[Kind]] is accessor.
VERIFY(value.value().is_accessor());
2023-04-14 17:05:02 +02:00
// 6. Let getter be entry.[[Get]].
auto* getter = value.value().as_accessor().getter();
2023-04-14 17:05:02 +02:00
// 5. If entry.[[Get]] is undefined, throw a TypeError exception.
if (!getter)
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldGetAccessorWithoutGetter, name.description);
2023-04-14 17:05:02 +02:00
// 7. Return ? Call(getter, O).
return TRY(call(vm, *getter, this));
}
2023-04-14 17:05:02 +02:00
// 7.3.32 PrivateSet ( O, P, value ), https://tc39.es/ecma262/#sec-privateset
ThrowCompletionOr<void> Object::private_set(PrivateName const& name, Value value)
{
auto& vm = this->vm();
2023-04-14 17:05:02 +02:00
// 1. Let entry be PrivateElementFind(O, P).
auto* entry = private_element_find(name);
2023-04-14 17:05:02 +02:00
// 2. If entry is empty, throw a TypeError exception.
if (!entry)
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldDoesNotExistOnObject, name.description);
2023-04-14 17:05:02 +02:00
// 3. If entry.[[Kind]] is field, then
if (entry->kind == PrivateElement::Kind::Field) {
2023-04-14 17:05:02 +02:00
// a. Set entry.[[Value]] to value.
entry->value = make_handle(value);
return {};
2023-04-14 17:05:02 +02:00
}
// 4. Else if entry.[[Kind]] is method, then
else if (entry->kind == PrivateElement::Kind::Method) {
// a. Throw a TypeError exception.
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldSetMethod, name.description);
}
2023-04-14 17:05:02 +02:00
// 5. Else,
// a. Assert: entry.[[Kind]] is accessor.
VERIFY(entry->kind == PrivateElement::Kind::Accessor);
auto& accessor = entry->value;
VERIFY(accessor.value().is_accessor());
2023-04-14 17:05:02 +02:00
// c. Let setter be entry.[[Set]].
auto* setter = accessor.value().as_accessor().setter();
2023-04-14 17:05:02 +02:00
// b. If entry.[[Set]] is undefined, throw a TypeError exception.
if (!setter)
return vm.throw_completion<TypeError>(ErrorType::PrivateFieldSetAccessorWithoutSetter, name.description);
2023-04-14 17:05:02 +02:00
// d. Perform ? Call(setter, O, « value »).
TRY(call(vm, *setter, this, value));
2023-04-14 17:05:02 +02:00
// 6. Return unused.
return {};
}
2023-04-14 17:05:02 +02:00
// 7.3.33 DefineField ( receiver, fieldRecord ), https://tc39.es/ecma262/#sec-definefield
ThrowCompletionOr<void> Object::define_field(ClassFieldDefinition const& field)
{
auto& vm = this->vm();
// 1. Let fieldName be fieldRecord.[[Name]].
auto const& field_name = field.name;
// 2. Let initializer be fieldRecord.[[Initializer]].
auto const& initializer = field.initializer;
auto init_value = js_undefined();
// 3. If initializer is not empty, then
if (!initializer.is_null()) {
// a. Let initValue be ? Call(initializer, receiver).
init_value = TRY(call(vm, initializer.cell(), this));
}
// 4. Else, let initValue be undefined.
// 5. If fieldName is a Private Name, then
if (field_name.has<PrivateName>()) {
// a. Perform ? PrivateFieldAdd(receiver, fieldName, initValue).
TRY(private_field_add(field_name.get<PrivateName>(), init_value));
}
// 6. Else,
else {
// a. Assert: IsPropertyKey(fieldName) is true.
// b. Perform ? CreateDataPropertyOrThrow(receiver, fieldName, initValue).
TRY(create_data_property_or_throw(field_name.get<PropertyKey>(), init_value));
}
// 7. Return unused.
return {};
}
2023-04-14 17:05:02 +02:00
// 7.3.34 InitializeInstanceElements ( O, constructor ), https://tc39.es/ecma262/#sec-initializeinstanceelements
ThrowCompletionOr<void> Object::initialize_instance_elements(ECMAScriptFunctionObject& constructor)
{
// 1. Let methods be the value of constructor.[[PrivateMethods]].
// 2. For each PrivateElement method of methods, do
for (auto const& method : constructor.private_methods()) {
// a. Perform ? PrivateMethodOrAccessorAdd(O, method).
TRY(private_method_or_accessor_add(method));
}
// 3. Let fields be the value of constructor.[[Fields]].
// 4. For each element fieldRecord of fields, do
for (auto const& field : constructor.fields()) {
// a. Perform ? DefineField(O, fieldRecord).
TRY(define_field(field));
}
// 5. Return unused.
return {};
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1 Ordinary Object Internal Methods and Internal Slots, https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots
// 10.1.1 [[GetPrototypeOf]] ( ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-getprototypeof
ThrowCompletionOr<Object*> Object::internal_get_prototype_of() const
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Return O.[[Prototype]].
return const_cast<Object*>(prototype());
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.2 [[SetPrototypeOf]] ( V ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-setprototypeof-v
ThrowCompletionOr<bool> Object::internal_set_prototype_of(Object* new_prototype)
{
// 1. Let current be O.[[Prototype]].
// 2. If SameValue(V, current) is true, return true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (prototype() == new_prototype)
return true;
// 3. Let extensible be O.[[Extensible]].
// 4. If extensible is false, return false.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!m_is_extensible)
return false;
// 5. Let p be V.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto* prototype = new_prototype;
// 6. Let done be false.
// 7. Repeat, while done is false,
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
while (prototype) {
// a. If p is null, set done to true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. Else if SameValue(p, O) is true, return false.
if (prototype == this)
return false;
// c. Else,
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// i. If p.[[GetPrototypeOf]] is not the ordinary object internal method defined in 10.1.1, set done to true.
// NOTE: This is a best-effort implementation; we don't have a good way of detecting whether certain virtual
// Object methods have been overridden by a given object, but as ProxyObject is the only one doing that for
// [[SetPrototypeOf]], this check does the trick.
if (is<ProxyObject>(prototype))
break;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// ii. Else, set p to p.[[Prototype]].
prototype = prototype->prototype();
}
// 8. Set O.[[Prototype]] to V.
set_prototype(new_prototype);
// 9. Return true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return true;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.3 [[IsExtensible]] ( ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-isextensible
ThrowCompletionOr<bool> Object::internal_is_extensible() const
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Return O.[[Extensible]].
return m_is_extensible;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.4 [[PreventExtensions]] ( ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-preventextensions
ThrowCompletionOr<bool> Object::internal_prevent_extensions()
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Set O.[[Extensible]] to false.
m_is_extensible = false;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. Return true.
return true;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.5 [[GetOwnProperty]] ( P ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-getownproperty-p
ThrowCompletionOr<Optional<PropertyDescriptor>> Object::internal_get_own_property(PropertyKey const& property_key) const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
// 1. If O does not have an own property with key P, return undefined.
auto maybe_storage_entry = storage_get(property_key);
if (!maybe_storage_entry.has_value())
return Optional<PropertyDescriptor> {};
// 2. Let D be a newly created Property Descriptor with no fields.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
PropertyDescriptor descriptor;
// 3. Let X be O's own property whose key is P.
auto [value, attributes, property_offset] = *maybe_storage_entry;
// 4. If X is a data property, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!value.is_accessor()) {
// a. Set D.[[Value]] to the value of X's [[Value]] attribute.
descriptor.value = value.value_or(js_undefined());
// b. Set D.[[Writable]] to the value of X's [[Writable]] attribute.
descriptor.writable = attributes.is_writable();
}
// 5. Else,
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
else {
// a. Assert: X is an accessor property.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. Set D.[[Get]] to the value of X's [[Get]] attribute.
descriptor.get = value.as_accessor().getter();
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// c. Set D.[[Set]] to the value of X's [[Set]] attribute.
descriptor.set = value.as_accessor().setter();
}
// 6. Set D.[[Enumerable]] to the value of X's [[Enumerable]] attribute.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
descriptor.enumerable = attributes.is_enumerable();
// 7. Set D.[[Configurable]] to the value of X's [[Configurable]] attribute.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
descriptor.configurable = attributes.is_configurable();
// Non-standard: Add the property offset to the descriptor. This is used to populate CacheablePropertyMetadata.
descriptor.property_offset = property_offset;
// 8. Return D.
return descriptor;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 10.1.6 [[DefineOwnProperty]] ( P, Desc ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-defineownproperty-p-desc
ThrowCompletionOr<bool> Object::internal_define_own_property(PropertyKey const& property_key, PropertyDescriptor const& property_descriptor)
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let current be ? O.[[GetOwnProperty]](P).
auto current = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. Let extensible be ? IsExtensible(O).
auto extensible = TRY(is_extensible());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. Return ValidateAndApplyPropertyDescriptor(O, P, extensible, Desc, current).
return validate_and_apply_property_descriptor(this, property_key, extensible, property_descriptor, current);
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.7 [[HasProperty]] ( P ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-hasproperty-p
ThrowCompletionOr<bool> Object::internal_has_property(PropertyKey const& property_key) const
{
VERIFY(property_key.is_valid());
// 1. Let hasOwn be ? O.[[GetOwnProperty]](P).
auto has_own = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If hasOwn is not undefined, return true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (has_own.has_value())
return true;
// 3. Let parent be ? O.[[GetPrototypeOf]]().
auto* parent = TRY(internal_get_prototype_of());
// 4. If parent is not null, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (parent) {
// a. Return ? parent.[[HasProperty]](P).
return parent->internal_has_property(property_key);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 5. Return false.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return false;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.8 [[Get]] ( P, Receiver ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-get-p-receiver
ThrowCompletionOr<Value> Object::internal_get(PropertyKey const& property_key, Value receiver, CacheablePropertyMetadata* cacheable_metadata) const
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
VERIFY(!receiver.is_empty());
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto& vm = this->vm();
// 1. Let desc be ? O.[[GetOwnProperty]](P).
auto descriptor = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. If desc is undefined, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!descriptor.has_value()) {
// a. Let parent be ? O.[[GetPrototypeOf]]().
auto* parent = TRY(internal_get_prototype_of());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. If parent is null, return undefined.
if (!parent)
return js_undefined();
// c. Return ? parent.[[Get]](P, Receiver).
return parent->internal_get(property_key, receiver);
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. If IsDataDescriptor(desc) is true, return desc.[[Value]].
if (descriptor->is_data_descriptor()) {
// Non-standard: If the caller has requested cacheable metadata and the property is an own property, fill it in.
if (cacheable_metadata && descriptor->property_offset.has_value()) {
*cacheable_metadata = CacheablePropertyMetadata {
.type = CacheablePropertyMetadata::Type::OwnProperty,
.property_offset = descriptor->property_offset.value(),
};
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return *descriptor->value;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 4. Assert: IsAccessorDescriptor(desc) is true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
VERIFY(descriptor->is_accessor_descriptor());
// 5. Let getter be desc.[[Get]].
auto getter = *descriptor->get;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 6. If getter is undefined, return undefined.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!getter)
return js_undefined();
// 7. Return ? Call(getter, Receiver).
return TRY(call(vm, *getter, receiver));
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.9 [[Set]] ( P, V, Receiver ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-set-p-v-receiver
ThrowCompletionOr<bool> Object::internal_set(PropertyKey const& property_key, Value value, Value receiver)
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
VERIFY(!value.is_empty());
VERIFY(!receiver.is_empty());
// 2. Let ownDesc be ? O.[[GetOwnProperty]](P).
auto own_descriptor = TRY(internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. Return ? OrdinarySetWithOwnDescriptor(O, P, V, Receiver, ownDesc).
return ordinary_set_with_own_descriptor(property_key, value, receiver, own_descriptor);
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.9.2 OrdinarySetWithOwnDescriptor ( O, P, V, Receiver, ownDesc ), https://tc39.es/ecma262/#sec-ordinarysetwithowndescriptor
ThrowCompletionOr<bool> Object::ordinary_set_with_own_descriptor(PropertyKey const& property_key, Value value, Value receiver, Optional<PropertyDescriptor> own_descriptor)
{
VERIFY(property_key.is_valid());
VERIFY(!value.is_empty());
VERIFY(!receiver.is_empty());
auto& vm = this->vm();
// 1. If ownDesc is undefined, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!own_descriptor.has_value()) {
// a. Let parent be ? O.[[GetPrototypeOf]]().
auto* parent = TRY(internal_get_prototype_of());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. If parent is not null, then
if (parent) {
// i. Return ? parent.[[Set]](P, V, Receiver).
return TRY(parent->internal_set(property_key, value, receiver));
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// c. Else,
else {
// i. Set ownDesc to the PropertyDescriptor { [[Value]]: undefined, [[Writable]]: true, [[Enumerable]]: true, [[Configurable]]: true }.
own_descriptor = PropertyDescriptor {
.value = js_undefined(),
.writable = true,
.enumerable = true,
.configurable = true,
};
}
}
// 2. If IsDataDescriptor(ownDesc) is true, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (own_descriptor->is_data_descriptor()) {
// a. If ownDesc.[[Writable]] is false, return false.
if (!*own_descriptor->writable)
return false;
// b. If Type(Receiver) is not Object, return false.
if (!receiver.is_object())
return false;
// c. Let existingDescriptor be ? Receiver.[[GetOwnProperty]](P).
auto existing_descriptor = TRY(receiver.as_object().internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// d. If existingDescriptor is not undefined, then
if (existing_descriptor.has_value()) {
// i. If IsAccessorDescriptor(existingDescriptor) is true, return false.
if (existing_descriptor->is_accessor_descriptor())
return false;
// ii. If existingDescriptor.[[Writable]] is false, return false.
if (!*existing_descriptor->writable)
return false;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// iii. Let valueDesc be the PropertyDescriptor { [[Value]]: V }.
auto value_descriptor = PropertyDescriptor { .value = value };
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// iv. Return ? Receiver.[[DefineOwnProperty]](P, valueDesc).
return TRY(receiver.as_object().internal_define_own_property(property_key, value_descriptor));
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// e. Else,
else {
// i. Assert: Receiver does not currently have a property P.
VERIFY(!receiver.as_object().storage_has(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// ii. Return ? CreateDataProperty(Receiver, P, V).
return TRY(receiver.as_object().create_data_property(property_key, value));
}
}
// 3. Assert: IsAccessorDescriptor(ownDesc) is true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
VERIFY(own_descriptor->is_accessor_descriptor());
// 4. Let setter be ownDesc.[[Set]].
auto setter = *own_descriptor->set;
// 5. If setter is undefined, return false.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!setter)
return false;
// 6. Perform ? Call(setter, Receiver, « V »).
(void)TRY(call(vm, *setter, receiver, value));
// 7. Return true.
return true;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.10 [[Delete]] ( P ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-delete-p
ThrowCompletionOr<bool> Object::internal_delete(PropertyKey const& property_key)
{
VERIFY(property_key.is_valid());
// 1. Let desc be ? O.[[GetOwnProperty]](P).
auto descriptor = TRY(internal_get_own_property(property_key));
// 2. If desc is undefined, return true.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!descriptor.has_value())
return true;
// 3. If desc.[[Configurable]] is true, then
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (*descriptor->configurable) {
// a. Remove the own property with name P from O.
storage_delete(property_key);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. Return true.
return true;
}
// 4. Return false.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return false;
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 10.1.11 [[OwnPropertyKeys]] ( ), https://tc39.es/ecma262/#sec-ordinary-object-internal-methods-and-internal-slots-ownpropertykeys
ThrowCompletionOr<MarkedVector<Value>> Object::internal_own_property_keys() const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
auto& vm = this->vm();
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let keys be a new empty List.
MarkedVector<Value> keys { heap() };
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. For each own property key P of O such that P is an array index, in ascending numeric index order, do
for (auto& entry : m_indexed_properties) {
// a. Add P as the last element of keys.
keys.append(PrimitiveString::create(vm, DeprecatedString::number(entry.index())));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. For each own property key P of O such that Type(P) is String and P is not an array index, in ascending chronological order of property creation, do
for (auto& it : shape().property_table()) {
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (it.key.is_string()) {
// a. Add P as the last element of keys.
keys.append(it.key.to_value(vm));
}
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 4. For each own property key P of O such that Type(P) is Symbol, in ascending chronological order of property creation, do
for (auto& it : shape().property_table()) {
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (it.key.is_symbol()) {
// a. Add P as the last element of keys.
keys.append(it.key.to_value(vm));
}
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 5. Return keys.
return { move(keys) };
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
// 10.4.7.2 SetImmutablePrototype ( O, V ), https://tc39.es/ecma262/#sec-set-immutable-prototype
ThrowCompletionOr<bool> Object::set_immutable_prototype(Object* prototype)
{
// 1. Let current be ? O.[[GetPrototypeOf]]().
auto* current = TRY(internal_get_prototype_of());
// 2. If SameValue(V, current) is true, return true.
if (prototype == current)
return true;
// 3. Return false.
return false;
}
static Optional<Object::IntrinsicAccessor> find_intrinsic_accessor(Object const* object, PropertyKey const& property_key)
{
if (!property_key.is_string())
return {};
auto intrinsics = s_intrinsics.find(object);
if (intrinsics == s_intrinsics.end())
return {};
auto accessor_iterator = intrinsics->value.find(property_key.as_string());
if (accessor_iterator == intrinsics->value.end())
return {};
auto accessor = accessor_iterator->value;
intrinsics->value.remove(accessor_iterator);
return accessor;
}
Optional<ValueAndAttributes> Object::storage_get(PropertyKey const& property_key) const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
Value value;
PropertyAttributes attributes;
Optional<u32> property_offset;
if (property_key.is_number()) {
auto value_and_attributes = m_indexed_properties.get(property_key.as_number());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!value_and_attributes.has_value())
return {};
value = value_and_attributes->value;
attributes = value_and_attributes->attributes;
} else {
auto metadata = shape().lookup(property_key.to_string_or_symbol());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!metadata.has_value())
return {};
if (m_has_intrinsic_accessors) {
if (auto accessor = find_intrinsic_accessor(this, property_key); accessor.has_value())
const_cast<Object&>(*this).m_storage[metadata->offset] = (*accessor)(shape().realm());
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
value = m_storage[metadata->offset];
attributes = metadata->attributes;
property_offset = metadata->offset;
}
return ValueAndAttributes { .value = value, .attributes = attributes, .property_offset = property_offset };
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
bool Object::storage_has(PropertyKey const& property_key) const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
VERIFY(property_key.is_valid());
if (property_key.is_number())
return m_indexed_properties.has_index(property_key.as_number());
return shape().lookup(property_key.to_string_or_symbol()).has_value();
}
void Object::storage_set(PropertyKey const& property_key, ValueAndAttributes const& value_and_attributes)
{
VERIFY(property_key.is_valid());
auto [value, attributes, _] = value_and_attributes;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (property_key.is_number()) {
auto index = property_key.as_number();
m_indexed_properties.put(index, value, attributes);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return;
}
if (m_has_intrinsic_accessors && property_key.is_string()) {
if (auto intrinsics = s_intrinsics.find(this); intrinsics != s_intrinsics.end())
intrinsics->value.remove(property_key.as_string());
}
auto property_key_string_or_symbol = property_key.to_string_or_symbol();
auto metadata = shape().lookup(property_key_string_or_symbol);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (!metadata.has_value()) {
if (!m_shape->is_unique() && shape().property_count() > 100) {
// If you add more than 100 properties to an object, let's stop doing
// transitions to avoid filling up the heap with shapes.
ensure_shape_is_unique();
}
if (m_shape->is_unique())
m_shape->add_property_to_unique_shape(property_key_string_or_symbol, attributes);
else
set_shape(*m_shape->create_put_transition(property_key_string_or_symbol, attributes));
m_storage.append(value);
return;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (attributes != metadata->attributes) {
if (m_shape->is_unique())
m_shape->reconfigure_property_in_unique_shape(property_key_string_or_symbol, attributes);
else
set_shape(*m_shape->create_configure_transition(property_key_string_or_symbol, attributes));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
}
m_storage[metadata->offset] = value;
}
void Object::storage_delete(PropertyKey const& property_key)
{
VERIFY(property_key.is_valid());
VERIFY(storage_has(property_key));
if (property_key.is_number())
return m_indexed_properties.remove(property_key.as_number());
if (m_has_intrinsic_accessors && property_key.is_string()) {
if (auto intrinsics = s_intrinsics.find(this); intrinsics != s_intrinsics.end())
intrinsics->value.remove(property_key.as_string());
}
auto metadata = shape().lookup(property_key.to_string_or_symbol());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
VERIFY(metadata.has_value());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
ensure_shape_is_unique();
shape().remove_property_from_unique_shape(property_key.to_string_or_symbol(), metadata->offset);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
m_storage.remove(metadata->offset);
}
void Object::set_prototype(Object* new_prototype)
{
if (prototype() == new_prototype)
return;
auto& shape = this->shape();
if (shape.is_unique())
shape.set_prototype_without_transition(new_prototype);
else
m_shape = shape.create_prototype_transition(new_prototype);
}
void Object::define_native_accessor(Realm& realm, PropertyKey const& property_key, Function<ThrowCompletionOr<Value>(VM&)> getter, Function<ThrowCompletionOr<Value>(VM&)> setter, PropertyAttributes attribute)
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
FunctionObject* getter_function = nullptr;
if (getter)
getter_function = NativeFunction::create(realm, move(getter), 0, property_key, &realm, {}, "get"sv);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
FunctionObject* setter_function = nullptr;
if (setter)
setter_function = NativeFunction::create(realm, move(setter), 1, property_key, &realm, {}, "set"sv);
return define_direct_accessor(property_key, getter_function, setter_function, attribute);
}
void Object::define_direct_accessor(PropertyKey const& property_key, FunctionObject* getter, FunctionObject* setter, PropertyAttributes attributes)
{
VERIFY(property_key.is_valid());
auto existing_property = storage_get(property_key).value_or({}).value;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
auto* accessor = existing_property.is_accessor() ? &existing_property.as_accessor() : nullptr;
if (!accessor) {
accessor = Accessor::create(vm(), getter, setter);
define_direct_property(property_key, accessor, attributes);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
} else {
if (getter)
accessor->set_getter(getter);
if (setter)
accessor->set_setter(setter);
}
}
void Object::define_intrinsic_accessor(PropertyKey const& property_key, PropertyAttributes attributes, IntrinsicAccessor accessor)
{
VERIFY(property_key.is_string());
storage_set(property_key, { {}, attributes });
m_has_intrinsic_accessors = true;
auto& intrinsics = s_intrinsics.ensure(this);
intrinsics.set(property_key.as_string(), move(accessor));
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
void Object::ensure_shape_is_unique()
{
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (shape().is_unique())
return;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
m_shape = m_shape->create_unique_clone();
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// Simple side-effect free property lookup, following the prototype chain. Non-standard.
2022-04-01 20:58:27 +03:00
Value Object::get_without_side_effects(PropertyKey const& property_key) const
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
{
auto* object = this;
while (object) {
auto value_and_attributes = object->storage_get(property_key);
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
if (value_and_attributes.has_value())
return value_and_attributes->value;
object = object->prototype();
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return {};
}
void Object::define_native_function(Realm& realm, PropertyKey const& property_key, Function<ThrowCompletionOr<Value>(VM&)> native_function, i32 length, PropertyAttributes attribute)
{
auto function = NativeFunction::create(realm, move(native_function), length, property_key, &realm);
define_direct_property(property_key, function, attribute);
}
// 20.1.2.3.1 ObjectDefineProperties ( O, Properties ), https://tc39.es/ecma262/#sec-objectdefineproperties
ThrowCompletionOr<Object*> Object::define_properties(Value properties)
{
auto& vm = this->vm();
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 1. Let props be ? ToObject(Properties).
auto props = TRY(properties.to_object(vm));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 2. Let keys be ? props.[[OwnPropertyKeys]]().
auto keys = TRY(props->internal_own_property_keys());
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
struct NameAndDescriptor {
PropertyKey name;
PropertyDescriptor descriptor;
};
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 3. Let descriptors be a new empty List.
Vector<NameAndDescriptor> descriptors;
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 4. For each element nextKey of keys, do
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
for (auto& next_key : keys) {
auto property_key = MUST(PropertyKey::from_value(vm, next_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// a. Let propDesc be ? props.[[GetOwnProperty]](nextKey).
auto property_descriptor = TRY(props->internal_get_own_property(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// b. If propDesc is not undefined and propDesc.[[Enumerable]] is true, then
if (property_descriptor.has_value() && *property_descriptor->enumerable) {
// i. Let descObj be ? Get(props, nextKey).
auto descriptor_object = TRY(props->get(property_key));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// ii. Let desc be ? ToPropertyDescriptor(descObj).
auto descriptor = TRY(to_property_descriptor(vm, descriptor_object));
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// iii. Append the pair (a two element List) consisting of nextKey and desc to the end of descriptors.
descriptors.append({ property_key, descriptor });
}
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 5. For each element pair of descriptors, do
for (auto& [name, descriptor] : descriptors) {
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// a. Let P be the first element of pair.
// b. Let desc be the second element of pair.
// c. Perform ? DefinePropertyOrThrow(O, P, desc).
TRY(define_property_or_throw(name, descriptor));
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 6. Return O.
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
return this;
}
// 14.7.5.9 EnumerateObjectProperties ( O ), https://tc39.es/ecma262/#sec-enumerate-object-properties
Optional<Completion> Object::enumerate_object_properties(Function<Optional<Completion>(Value)> callback) const
{
// 1. Return an Iterator object (27.1.1.2) whose next method iterates over all the String-valued keys of enumerable properties of O. The iterator object is never directly accessible to ECMAScript code. The mechanics and order of enumerating the properties is not specified but must conform to the rules specified below.
// * Returned property keys do not include keys that are Symbols.
// * Properties of the target object may be deleted during enumeration.
// * A property that is deleted before it is processed is ignored.
// * If new properties are added to the target object during enumeration, the newly added properties are not guaranteed to be processed in the active enumeration.
// * A property name will be returned at most once in any enumeration.
// * Enumerating the properties of the target object includes enumerating properties of its prototype, and the prototype of the prototype, and so on, recursively.
// * A property of a prototype is not processed if it has the same name as a property that has already been processed.
HashTable<DeprecatedFlyString> visited;
auto const* target = this;
while (target) {
auto own_keys = TRY(target->internal_own_property_keys());
for (auto& key : own_keys) {
if (!key.is_string())
continue;
DeprecatedFlyString property_key = key.as_string().deprecated_string();
if (visited.contains(property_key))
continue;
auto descriptor = TRY(target->internal_get_own_property(property_key));
if (!descriptor.has_value())
continue;
visited.set(property_key);
if (!*descriptor->enumerable)
continue;
if (auto completion = callback(key); completion.has_value())
return completion.release_value();
}
target = TRY(target->internal_get_prototype_of());
};
return {};
}
void Object::visit_edges(Cell::Visitor& visitor)
{
Base::visit_edges(visitor);
visitor.visit(m_shape);
for (auto& value : m_storage)
visitor.visit(value);
m_indexed_properties.for_each_value([&visitor](auto& value) {
visitor.visit(value);
});
}
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
// 7.1.1.1 OrdinaryToPrimitive ( O, hint ), https://tc39.es/ecma262/#sec-ordinarytoprimitive
ThrowCompletionOr<Value> Object::ordinary_to_primitive(Value::PreferredType preferred_type) const
{
VERIFY(preferred_type == Value::PreferredType::String || preferred_type == Value::PreferredType::Number);
auto& vm = this->vm();
AK::Array<PropertyKey, 2> method_names;
// 1. If hint is string, then
if (preferred_type == Value::PreferredType::String) {
// a. Let methodNames be « "toString", "valueOf" ».
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
method_names = { vm.names.toString, vm.names.valueOf };
}
// 2. Else,
else {
// a. Let methodNames be « "valueOf", "toString" ».
LibJS: Rewrite most of Object for spec compliance :^) This is a huge patch, I know. In hindsight this perhaps could've been done slightly more incremental, but I started and then fixed everything until it worked, and here we are. I tried splitting of some completely unrelated changes into separate commits, however. Anyway. This is a rewrite of most of Object, and by extension large parts of Array, Proxy, Reflect, String, TypedArray, and some other things. What we already had worked fine for about 90% of things, but getting the last 10% right proved to be increasingly difficult with the current code that sort of grew organically and is only very loosely based on the spec - this became especially obvious when we started fixing a large number of test262 failures. Key changes include: - 1:1 matching function names and parameters of all object-related functions, to avoid ambiguity. Previously we had things like put(), which the spec doesn't have - as a result it wasn't always clear which need to be used. - Better separation between object abstract operations and internal methods - the former are always the same, the latter can be overridden (and are therefore virtual). The internal methods (i.e. [[Foo]] in the spec) are now prefixed with 'internal_' for clarity - again, it was previously not always clear which AO a certain method represents, get() could've been both Get and [[Get]] (I don't know which one it was closer to right now). Note that some of the old names have been kept until all code relying on them is updated, but they are now simple wrappers around the closest matching standard abstract operation. - Simplifications of the storage layer: functions that write values to storage are now prefixed with 'storage_' to make their purpose clear, and as they are not part of the spec they should not contain any steps specified by it. Much functionality is now covered by the layers above it and was removed (e.g. handling of accessors, attribute checks). - PropertyAttributes has been greatly simplified, and is being replaced by PropertyDescriptor - a concept similar to the current implementation, but more aligned with the actual spec. See the commit message of the previous commit where it was introduced for details. - As a bonus, and since I had to look at the spec a whole lot anyway, I introduced more inline comments with the exact steps from the spec - this makes it super easy to verify correctness. - East-const all the things. As a result of all of this, things are much more correct but a bit slower now. Retaining speed wasn't a consideration at all, I have done no profiling of the new code - there might be low hanging fruits, which we can then harvest separately. Special thanks to Idan for helping me with this by tracking down bugs, updating everything outside of LibJS to work with these changes (LibWeb, Spreadsheet, HackStudio), as well as providing countless patches to fix regressions I introduced - there still are very few (we got it down to 5), but we also get many new passing test262 tests in return. :^) Co-authored-by: Idan Horowitz <idan.horowitz@gmail.com>
2021-07-04 18:14:16 +01:00
method_names = { vm.names.valueOf, vm.names.toString };
}
// 3. For each element name of methodNames, do
for (auto& method_name : method_names) {
// a. Let method be ? Get(O, name).
auto method = TRY(get(method_name));
// b. If IsCallable(method) is true, then
if (method.is_function()) {
// i. Let result be ? Call(method, O).
auto result = TRY(call(vm, method.as_function(), const_cast<Object*>(this)));
// ii. If Type(result) is not Object, return result.
if (!result.is_object())
return result;
}
}
// 4. Throw a TypeError exception.
return vm.throw_completion<TypeError>(ErrorType::Convert, "object", preferred_type == Value::PreferredType::String ? "string" : "number");
}
}