Our floating point number parser was based on the fast_float library:
https://github.com/fastfloat/fast_float
However, our implementation only supports 8-bit characters. To support
UTF-16, we will need to be able to convert char16_t-based strings to
numbers as well. This works out-of-the-box with fast_float.
We can also use fast_float for integer parsing.
By definition, the web allows lonely surrogates by default. Let's have
our string APIs reflect this, so we don't have to pass an allow option
all over the place.
To prepare for an upcoming Utf16String, this migrates Utf16View to store
its data as a char16_t. Most function definitions are moved inline and
made constexpr.
This also adds a UDL to construct a Utf16View from a string literal:
auto string = u"hello"sv;
This let's us remove the NTTP Utf16View constructor, as we have found
that such constructors bloat binary size quite a bit.
`TypedArray` objects need to know their own constructor objects to allow
copying. This was implemented by storing a function pointer to the
`Intrinsic` object's method which returns the constructor object.
The problem is that function pointers aren't polymorphic, we can't
legally just cast e.g. a `Derived* (*ptr)(void)` to `Base*
(*ptr)(void)` (this is why the code needed a `bit_cast` to even
compile). But this wasn't actually a problem in practice because their
ABIs were the same. But with pointer authentication (Apple's `arm64e`
ABI) this signature mismatch becomes a hard failure and crashes the
process.
Fix this by adding a virtual function that returns the intrinsic
constructor (actually, a `NativeFunction`, as typed arrays constructors
don't inherit from the base `TypedArray` constructor) instead of the
function pointer.
With this, test-js passes and Ladybird launches correctly when built
(with a lot of vcpkg hacks) for arm64e.
Instead, we can just use the scope type to determine if a scope is a
function scope.
This fixes using `this` for parameter default values in arrow functions
crashing. This happened by `uses_this_from_environment` was not set in
`set_uses_this`, as it didn't think it was in a function scope whilst
parsing parameters.
Fixes closing modal dialogs causing a crash on https://www.ikea.com/
No test262 diff.
Reverts the functional part of 08cfd5f, because it was a workaround for
this issue.
...of Array. If array has simple storage, which implies that attributes
of all indexed properties are default, and newly added property also
has default attribute, we can do a fast path and skip lots of checks
that happen in `Object::internal_define_own_property()`.
If array has packed index property storage without holes, we could check
if indexed property is present simple by checking if it's less than
array's length.
Makes the following program go 1.1x faster:
```js
function f() {
let array = [];
for (let i = 0; i < 3_000; i++) {
array.push(i);
}
for (let i = 0; i < 10_000; i++) {
array.map(x => x * 2);
}
}
f();
```
This one is required to cover the case when new empty elements are
introduced by assigning to element with index > length, like:
```js
var x = [];
x[0] = 1;
x[2] = 2;
```
...by avoiding `CreateListFromArrayLike` in cases when we could directly
use elements of underlying object's indexed properties storage.
Makes this program go 2.1x faster:
```js
function target(a, b, c) {
return a + b + c;
}
const args = [1, 2, 3];
let result = 0;
(function() {
for (let i = 0; i < 10_000_000; i++) {
result += target.apply(null, args);
}
})();
```
Before this change each built-in iterator object has a boolean
`m_next_method_was_redefined`. If user code later changed the iterator’s
prototype (e.g. `Object.setPrototypeOf()`), we still believed the
built-in fast-path was safe and skipped the user supplied override,
producing wrong results.
With this change
`BuiltinIterator::as_builtin_iterator_if_next_is_not_redefined()` looks
up the current `next` property and verifies that it is still the
built-in native function.
This mirrors the existing caching logic for int32 constants.
Avoids duplication of string constants in m_constants which could
result in stack overflows for large scripts with a lot of similar
strings.
This commit adds the minimal export macros needed to run js.exe on
windows. A followup commit is planned to move to explicit export
entirely.
A static_assert for the size of a struct is also ifdef'ed out as the
semantics around object layout and inheritance are different on MSVC abi
and the struct IteratorRecord ends up being 40 bytes not 32.
In the following example:
```js
const f = (i) => ({
obj: { a: { x: i }, b: { x: i } },
g: () => {},
});
```
The body of function `f` is initially parsed as an arrow function. As a
result, what is actually an object expression is interpreted as a formal
parameter with a binding pattern. Since duplicate identifiers are not
allowed in this context (`i` in the example), the parser generates an
error, causing the entire script to fail parsing.
This change ignores the "Duplicate parameter names in bindings" error
during arrow function parameter parsing, allowing the parser to continue
and recognize the object expression of the outer arrow function with an
implicit return.
Fixes error on https://chat.openai.com/
...when Array.prototype and Object.prototype are intact.
If `internal_set()` is called on an array exotic object with a numeric
PropertyKey, and:
- the prototype chain has not been modified (i.e., there are no getters
or setters for indexed properties), and
- the array is not the target of a Proxy object,
then we can directly store the value in the receiver's indexed
properties, without checking whether it already exists somewhere in the
prototype chain.
1.7x improvement on the following program:
```js
function f() {
let a = [];
let i = 0;
while (i < 10_000_000) {
a.push(i);
i++;
}
}
f();
```
Replace the implementation of maths in `UnsignedBigInteger`
and `SignedBigInteger` with LibTomMath. This gives benefits in terms of
less code to maintain, correctness and speed.
These changes also remove now-unsued methods and improve the error
propagation for functions allocating lots of memory. Additionally, the
new implementation is always trimmed and won't have dangling zeros when
exporting it.
Having it as a method instead of a free function is necessary for the
next commits and generally allows for optimizations that require deeper
access into the `UnsignedBigInteger` / `SignedBigInteger`.
Also restrict the exponent to 32 bits to avoid huge memory allocations.
This is functionaly the same since caller_context is the topmost
execution context on the stack, but makes it more clear that
we are directly inheriting the strict mode from the caller context
when pushing the next context on to the stack.
