This is a relatively small code size regression (130k on windows and
macOS, 180k on Linux), for a few high confidence improvements in
speedometer 3. See link in the bug.
If this sticks, we can actually clean up a bunch of code, and eventually
unify RefPtr and nsCOMPtr. But I want this to be on the tree for a while
before doing more aggressive refactorings / actually removing the code.
Differential Revision: https://phabricator.services.mozilla.com/D178267
This changes the cssparser setup to:
* Avoid having to do copies of the ParsingContext all over the place,
which is useful because I plan to stash more nesting state in there.
* Use the new RuleBodyParser which allows parsing qualified rules,
declarations, and so on. Though we still don't use this anywhere.
The next step is to join NestedRuleParser and PropertyDeclarationParser,
so that we can parse declarations in a lot of the nested rules as well.
Differential Revision: https://phabricator.services.mozilla.com/D178053
Unlike the `ParseRelative` flag, which turns on relative selector parsing for
that parsing level only, the newly added `DISALLOW_RELATIVE_SELECTOR` flag
propagates through nested parsing calls.
Differential Revision: https://phabricator.services.mozilla.com/D176807
This simplifies a tiny bit our bindings in some places, and complicates
it in others, but over all I think it's better.
It requires a bit more manual code in the rust side to drop and cast the
relevant pointers (which was done implicitly before), but it's a lot
less magic than it used to be, and that's all autogenerated so consumers
don't need to care about it.
The set up is still not ideal. I don't like that we rely on destructors
running in both sides of the FFI boundary, but that's for another day.
This is the last usage of RawOffsetArc, so remove that. We now support
proper Arc<> in structs (GridTemplateAreas uses it), so I don't think
we'll need it any time soon.
Differential Revision: https://phabricator.services.mozilla.com/D177905
That's how we do it for all other platforms. Do this rather than via a
custom mostly-untested pref, which allows us to simplify text zoom
handling.
Differential Revision: https://phabricator.services.mozilla.com/D177062
Much like with font-relative units, when they're used for a
non-inherited property we need to avoid using the rule cache.
This is because two elements matching the same rules won't get
guaranteed-equal non-inherited style structs.
Depends on D177732
Differential Revision: https://phabricator.services.mozilla.com/D177733
See blocked bug.
For non-Locked types we can just use the same underlying type at the FFI
boundary. Port StylesheetContents and CssUrlData to this set-up.
CssUrlData is already generated by cbindgen anyways.
Differential Revision: https://phabricator.services.mozilla.com/D177559
After the previous patches we only have one trait which we should also
tweak / rework, so let's put it all on that single trait.
Differential Revision: https://phabricator.services.mozilla.com/D177515
HasBoxFFI and HasArcFFI aren't great, see bug 1831242 as for examples of
why.
HasArcFFI requires a bit more care, but HasBoxFFI doesn't give us much
benefit. Instead use the same type in the FFI boundary.
Differential Revision: https://phabricator.services.mozilla.com/D177252
This restores the pre-regression behavior by choosing the later class in
cases where folks use stuff like `.foo.bar`.
This matches other browsers too.
Differential Revision: https://phabricator.services.mozilla.com/D177398
`self` keyword specifies to use the element’s own principal box as the
scroll container. If the principal box is not a scroll container, then the
scroll progress timeline is inactive.
Differential Revision: https://phabricator.services.mozilla.com/D175707
The order of <scroller> and <axis> doesn't matter in the parser.
However, we serialize <scroller> first, if it is not the initial value.
Differential Revision: https://phabricator.services.mozilla.com/D173906
And use it instead of explicit document checks. This centralizes where
we check for it.
IsChromeDoc is relatively cheap, but this bug wants to also check for
PDF.js which is a bit more expensive.
No behavior change.
Differential Revision: https://phabricator.services.mozilla.com/D176940
Sum nodes would use mul_by to negate nodes to do subtraction, but some
nodes are not distributive. This patch adds a negate node, so that the
operations inside these negate nodes can be resolved first and then the
"subtraction" can be applied.
Differential Revision: https://phabricator.services.mozilla.com/D172941
Support view() notation for animation-timeline:
`<view()> = view( [ <axis> || <'view-timeline-inset'> ]? )`
We move AnimationTimeline and its related types into the generics folder,
and define two new structs for scroll() and view().
Note:
1. The syntax of scroll() doesn't match the current version of the spec.
I will update it in Bug 1814444.
2. We will handle the creation/usage of the Anonymous View Progress Timelines
in the next patch.
Differential Revision: https://phabricator.services.mozilla.com/D173904
Although we store animation and transition style values in StyleUIReset and
define their properties in longhands/ui.mako.rs, but we may move them in
the future if this style struct becomes too large. So let's move the
definition of their values to an independent module, animation, so we
don't have to worry about this again.
This patch doesn't change any other things. Only move code.
Differential Revision: https://phabricator.services.mozilla.com/D173903
NaN and infinity percentages are now serialized as expected.
Also added some new WPT tests as percentages were previously untested
and added some spec comments to previous NaN/inf serialization code.
Differential Revision: https://phabricator.services.mozilla.com/D176726
No implementation just yet (the default QualifiedRuleParser
implementation just rejects stuff), but this is plumbing that I'd rather
get reviewed separately.
Differential Revision: https://phabricator.services.mozilla.com/D176686
More nesting plumbing. Still does nothing because we don't parse the
nested rules.
Should be trivial to prove this patch doesn't change any behavior so
far, but I want to land it on its own because it can have performance
implications.
This follows the pattern of what we do with other rules like layers and
container conditions, that is, keep the ancestor selectors in a stack,
and poke at the last one in order to replace the ancestor.
This changes the behavior of replace_parent_selector as with the newer
version of the spec, stuff like:
div {
.foo {
stuff
}
}
Should work as `div .foo`. A test is added for this case.
Differential Revision: https://phabricator.services.mozilla.com/D176560
