Vectors that stick around in the AST were wasting a fair bit of memory
due to the growth padding we keep by default. This patch goes after some
of these vectors with the shrink_to_fit() stick to reduce waste.
Since the AST can stay around for a long time, it is worth making an
effort to shrink it down when we have a chance.
After setting up all the bindings in function_declaration_instantiation,
we now ask DeclarativeEnvironment to do a shrink_to_fit() on its vector
of bindings.
This ends up saving 5.6 MiB on twitter.com/awesomekling :^)
Instead of CallExpression storing its arguments in a Vector<Argument>,
we now custom-allocate the memory slot for CallExpression (and its
subclass NewExpression) so that it fits both CallExpression and its list
of Arguments in one allocation.
This reduces memory usage on twitter.com/awesomekling by 8.8 MiB :^)
This template allows us to allocate an AST node and an array of some
arbitrary type T with one allocation instead of two. This can save
a lot of memory in some cases.
Thanks to Jonathan Müller for suggesting this technique! :^)
ASTNode inherits from RefCounted, which has a single 32-bit member.
This means that there's a 32-bit padding hole after RefCounted,
where we are free to put something (or it will go to waste!)
This patch moves ASTNode::m_start_offset into that padding hole,
and we now have a 32-bit padding hole at the end of ASTNode instead.
This will allow ASTNode subclasses to put things in the ASTNode hole
by moving them to the head of the member list.
Similar to how LibJS and LibSQL used to behave, the boolean constructor
of JsonValue is currently allowing pointers to be used to construct a
boolean value. Explicitly disallow such construction.
The getters passed to REGISTER_STRING_PROPERTY are never invoked. But if
they were, they would errantly incur an implicit pointer-to-boolean cast
when their return type (Gfx::Bitmap*) is used to construct a JsonValue.
Widget's name are the current way to retrieve them when using GML.
Presently, there is no way to differentiate two items that share the
same name.
`IncrementalSearchBanner` uses common names as "close_button" or
"next_button", prepend them with `incremental_search_banner_` avoid
collisions.
This fixes a bug where the close button of `CrashReporter` was confused
with the one of the search banner.
However, This solution isn't perfect, down the road, we should probably
find a way to warn about equal names and introduce something like
namespace to avoid huge prefixes.
Currently, cookies are ephemeral and only survive for the lifetime of
Browser instance. This will make Browser instead store cookies in a SQL
database for persisted access.
Updating cookies through these hooks happens in one of two manners:
1. Through the Browser's storage inspector.
2. Through WebDriver's delete-cookies operation.
In (1), we should not restrict ourselves to being able to delete cookies
for the current page. For example, it's handy to open the inspector from
the welcome page and be able to delete cookies for any domain.
In (2), we already are only interacting with cookies that have been
matched against the document URL.
This also allows for overriding the path. Ladybird will want to store
the database files in a subdirectory of the standard data directory that
contains the Ladybird application name.
Fixes#16000.
PNGWriter sets up one dummy scanline with the same width as the other
scanlines in order to allow addressing the "previous scanline" without
complicating the code.
By using a FixedArray instead of a VLA, we sidestep the risk of stack
overflow and instead get something that can signal OOM.
This patch basically uses the TRY() macro throughout PNGWriter instead
of relying on the MUST()'ing wrappers in Vector and ByteBuffer.
One FIXME was killed in the making of this patch. :^)
This is a first step towards handling PNG encoding failures instead of
just falling over and crashing the program.
This initial step will cause encode() to return an error if the final
ByteBuffer copy fails to allocate. There are more potential failures
that will be surfaced by subsequent commits.
Two FIXMEs were killed in the making of this patch. :^)
`Core::Stream::File` shouldn't hold any utility methods that are
unrelated to constructing a `Core::Stream`, so let's just replace the
existing `Core::File::exists` with the nicer looking implementation.
These are an attempt to separate the internal "pixel" used by CSS from
the actual "pixel" that exists on the display. Because of things like
2x display scaling, the ratio between these can vary, so having
distinct types will help prevent errors when converting from one unit
to the other.
`CSSPixels` refers to the `px` unit used on the web, which depending on
the device may or may not map to 1 pixel on the physical display. It's
a wrapper around `float`, and will be used by LibWeb for size and
position values up until we go to paint them to the screen.
`DevicePixels` on the other hand is a 1-to-1 pixel on the physical
display. It's a wrapper around `int`.
Without this change, the upcoming LibWeb pixel types will require a
silly doubled conversion in some places.
eg: `some_rect.to_type<int>().to_type<float>()`
With these overloads, we can get away with `some_rect.to_type<int>()`.
Before this commit it was a bit ambiguous which buttons the function
name were referring to; this instead now makes it clear that it's
related to mouse input. Additionally, this also fixes incorrect getter
naming leftover from yesteryear.
