The dependency count is not at all minor, and this way we avoid looping
through all of them in the common cases, mainly either changing state, or
attributes.
The style candidate cache had regressed a few times (see #12534), and my
intuition is that being able to disable all style sharing with a single rule in
the page is really unfortunate.
This commit redesigns the style sharing cache in order to be a optimistic cache,
but then reject candidates if they match different sibling-affecting selectors
in the page, for example.
So far the numbers have improved, but not so much as I'd wanted (~10%/20% of
non-incremental restyling time in general). The current implementation is really
dumb though (we recompute and re-match a lot of stuff), so we should be able to
optimise it quite a bit.
I have different ideas for improving it (that may or may not work), apart of the
low-hanging fruit like don't re-matching candidates all the time but I have to
measure the real impact.
Also, I need to verify it against try.
This is a rewrite for how style interfaces with its consumers in order to allow
different representations for an element snapshot.
This also changes the requirements of an element snapshot, requiring them to
only implement MatchAttr, instead of MatchAttrGeneric. This is important for
stylo since implementing MatchAttrGeneric is way more difficult for us given the
atom limitations. This also allows for more performant implementations in the
Gecko side of things.
This commit refactors the style crate to be completely independent of
the actual implementation and pseudo-elements supported.
This also adds a gecko backend which introduces parsing for the
anonymous box pseudo-elements[1], although there's still no way of
querying them.
https://mxr.mozilla.org/mozilla-central/source/layout/style/nsCSSAnonBoxList.h
Updated string_cache, html5ever, xml5ever and selectors in Cargo.toml files and Cargo.lock.
Removed references to string_cache_plugin.
Import atom! and ns! from string_cache.
Replaced ns!("") by ns!().
Replaced ns!(XML) and co by ns!(xml) and co.
Replaced atom!(foo) by atom!("foo").
Replaced Atom::from_slice by Atom::from.
Replaced atom.as_slice() by &*atom.
This is the strategy we'll need to take for attributes, and so this change
puts us in a position to handle attributes and state the same way.
This does mean that we stop taking care to track the situations where our
state has reverted to the original state, with no net change. I think that's
probably of negligible value though.
For some reason when I wrote this code I mixed up the rules for pseudo-elements
(rightmost, maximum of one) with those of pseudo-classes (which have no such
restriction). This caused a correctness issue which can be demonstrated with the
associated reftest modification.
