style: Avoid scanning the longhand id twice in the fast path of PropertyDeclarationBlock::parse_common
This makes the overhead of this function much lower according to `perf`.
And in general makes sense, removing overhead from the fast path.
<!-- Reviewable:start -->
---
This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/18281)
<!-- Reviewable:end -->
This makes the overhead of this function much lower according to `perf`.
And in general makes sense, removing overhead from the fast path.
MozReview-Commit-ID: 4dJRap9dxNG
Signed-off-by: Emilio Cobos Álvarez <emilio@crisal.io>
Devirtualize CSS error reporting.
This removes a trait object from the path of reporting a CSS error.
---
- [X] `./mach build -d` does not report any errors
- [X] `./mach test-tidy` does not report any errors
- [X] There are tests for these changes
<!-- Reviewable:start -->
---
This change is [<img src="https://reviewable.io/review_button.svg" height="34" align="absmiddle" alt="Reviewable"/>](https://reviewable.io/reviews/servo/servo/18209)
<!-- Reviewable:end -->
This is a temporary step needed to support Gecko's getKeyframes() API
until we implement bug 1391537. It only takes effect when
a ComputedValues object is supplied and only for longhand declarations.
This is unfortunate, but for now it complicates things, I would like not needing
a ElementData to get the style of an Element in order to fix all the
getDefaultComputedStyle bugs.
MozReview-Commit-ID: LZvsdFEqrDE
In the next few patches we move all non-transition related code over to
using AnimatableLonghand instead of TransitionProperty. This will allow
us to re-purpose TransitionProperty to represent only properties that
can be transitioned (i.e. excluding discrete properties) as well as
simplifying the code by removing the need to deal with shorthands and
the "all" value in places that do not need to handle those values.
Concretely we avoid allocating and scanning a temporary vector of longhands not
yet serialized for each shorthand.
This doesn't save a lot of time in Linux, but I bet it's somewhat important on
OSX.
At least when the longhands aren't custom properties.
We should also look into not serializing the style attribute eagerly when it's
not needed... But a lot of code currently rely on attribute values being
dereferenciables to &str, so that's harder to fix.
We should really look into all those vectors around too, but that's probably
less urgent.
The CSSOM spec forces us to reuse the slot, so we need to conditionally keep the
previous behavior.
This fixes bug 1369198.
MozReview-Commit-ID: LM9mK6ngZ4e
MallocSizeOf is similar to the existing HeapSizeOf trait from the
heapsize crate. The only difference is that MallocSizeOf's
malloc_size_of_children() function takes an additional MallocSizeOfFn
argument, which is used to measure heap blocks. This extra argument
makes MallocSizeOf match how Gecko's memory measurements work, and is
required for Stylo to integrate with DMD.
The patch also introduces a second trait, MallocSizeOfWithGuard, which
is much the same as MallocSizeOf, but with a |guard| argument for the
global style lock.
Finally, the patch uses the new traits to measure a small amount of
Stylo's memory usage.
This enum type used to contain the result of parsing
one CSS source declaration (`name: value;`) and expanding shorthands.
Enum types are as big as the biggest of their variant (plus discriminant),
which was quite big because some shorthands
expand to many longhand properties.
This type was returned through many functions and methods,
wrapped and rewrapped in `Result` with different error types.
This presumably caused significant `memmove` traffic.
Instead, we now allocate an `ArrayVec` on the stack
and pass `&mut` references to it for various functions to push into it.
This type is also very big, but we never move it.
We still use an intermediate data structure because we sometimes decide
after shorthand expansion that a declaration is invalid after all
and that we’re gonna drop it.
Only later do we push to a `PropertyDeclarationBlock`,
with an entire `ArrayVec` or nothing.
In future work we can try to avoid a large stack-allocated array,
and instead writing directly to the heap allocation
of the `Vec` inside `PropertyDeclarationBlock`.
However this is tricky:
we need to preserve this "all or nothing" aspect
of parsing one source declaration,
and at the same time we want to make it as little error-prone as possible
for the various call sites.
`PropertyDeclarationBlock` curently does property deduplication
incrementally: as each `PropertyDeclaration` is pushed,
we check if an existing declaration of the same property exists
and if so overwrite it.
To get rid of the stack allocated array we’d need to somehow
deduplicate separately after pushing multiple `PropertyDeclaration`.
We need another flag that represents allow-negative-number for SMIL, so
this enum will also comprise the another parsing mode that allows negative number.
Now that the `context` contains the `rule_type`, we can remove the `in_keyframe`
arg and check the `rule_type` to achieve the same thing.
MozReview-Commit-ID: oXrFBPuKMz
Extend Servo's @page parsing to match Gecko's CSS 2.2 behavior, where only
margin properties are allowed in an @page rule. Other properties are ignored.
MozReview-Commit-ID: IPYUlnkLYSb
