This is probably an old-ish bug made more frequent by the font loading
optimizations.
PostRebuildAllStyleData is a bit of a misnomer, but was always calling
ClearCachedData() on the style set, even if we weren't guaranteed to restyle
every element.
This means both wasted work and correctness issues (as the "uses <rare-feature>"
bits are cleared during this call, on the assumption that we'll then visit all
elements and that'd recompute it properly).
For now, unify a bit the different code paths and only clear these bits if we're
guaranteed to restyle all elements.
I should rename this to something better in a follow-up, and ideally also
decouple the ClearCachedData() calls a bit...
Differential Revision: https://phabricator.services.mozilla.com/D65740
--HG--
extra : moz-landing-system : lando
This fixes another edge-case that I thought of while debugging this, I think
this makes our behavior correct now. The comment and test-case should be
self-descriptive.
Differential Revision: https://phabricator.services.mozilla.com/D48135
--HG--
extra : moz-landing-system : lando
One of the issues with the test-case in this bug is that the page consumes a ton
of CPU due to scroll anchor adjustments being triggered from scroll events,
which in turn cause other scroll events to fire.
This happens in Chrome as well (just scroll to the bottom of the test-case, and
do `addEventListener("scroll", () => console.log("scroll"))` on devtools. But I
think it's worth fixing. This patch fixes it and overall I think it's a slightly
better approach to suppress adjustments than what we're doing.
Differential Revision: https://phabricator.services.mozilla.com/D39339
--HG--
extra : moz-landing-system : lando
This is IMO the right RestyleManager change for what bug 1527210 tried to fix.
We need to apply the animation hints to the primary frame, not the style frame.
The other non-RestyleManager bits of that bug still apply and look fine to me.
Differential Revision: https://phabricator.services.mozilla.com/D38598
--HG--
extra : moz-landing-system : lando
`*Inlines.h` shouldn't be included by another header file, but `nsPresContext.h`
does it. This causes include-hell which blocks the following fix.
Additionally, it causes an include hell between `PresShell.h` vs.
`nsIPresShell.h` and `nsPresContext.h if `Document.h` includes `PresShell.h`.
Therefore, this patch also solves this include hell with adding
`nsPresContextInlines.h`.
Differential Revision: https://phabricator.services.mozilla.com/D25333
--HG--
extra : moz-landing-system : lando
`*Inlines.h` shouldn't be included by another header file, but `nsPresContext.h`
does it. This causes include-hell which blocks the following fix.
Additionally, it causes an include hell between `PresShell.h` vs.
`nsIPresShell.h` and `nsPresContext.h if `Document.h` includes `PresShell.h`.
Therefore, this patch also solves this include hell with adding
`nsPresContextInlines.h`.
Differential Revision: https://phabricator.services.mozilla.com/D25333
--HG--
extra : moz-landing-system : lando
This doesn't matter yet because all the states that return a change hint are on
stylesheets, but will matter with bug 1472637.
Differential Revision: https://phabricator.services.mozilla.com/D21616
This doesn't matter yet because all the states that return a change hint are on
stylesheets, but will matter with bug 1472637.
Differential Revision: https://phabricator.services.mozilla.com/D21616
--HG--
extra : moz-landing-system : lando
This is more consistent with what the Rust bits of the style system do, and
removes a pointer from ComputedStyle which is always nice.
This also aligns the Rust bits with the C++ bits re. not treating xul pseudos as
anonymous boxes. See the comment in nsTreeStyleCache.cpp regarding those.
Can't wait for XUL trees to die.
Depends on D19001
Differential Revision: https://phabricator.services.mozilla.com/D19002
--HG--
extra : moz-landing-system : lando
The applying change hints are the same as what we apply for transform style
changed from something to 'none'.
All test cases pass with this fix fail without the fix.
MozReview-Commit-ID: 7HStU26lRPq
--HG--
extra : rebase_source : 701361d477ab7954ab15fde75fd40d98724e47b3
Now that BeginUpdate is useless for the UPDATE_STYLE case, we don't need the
update mechanism at all. Just ensure that ApplicableStylesChanged is called on
the pres shell via the relevant RuleChanged, etc. notifications.
There's a big hidden gotcha here. nsIDocument::BeginUpdate does put a script
blocker on the stack for these updates. However it's not needed, since no script
can run during these notifications (only the stylesheet events we post for
devtools, but those use AsyncEventDispatcher and PostDOMEvents, so they don't
try to run immediately).
nsIDocument::BeginUpdate also does XBL binding attached queue stuff, but we
can't change bindings during these notifications anyway, so it also doesn't
matter.
MozReview-Commit-ID: HJvK6zQfloh
Much in the spirit of bug 1442207.
They're not only unneeded, and cheap to get, but also we call them
inconsistently with the light DOM and flattened tree parent (like ContentRemoved
for display: contents), so they're really confusing, and kind of a footgun.
MozReview-Commit-ID: 9u3Kp8Kpp5i
This unfortunately doesn't fix my test-case (because we're replacing the text
content all the time and all that), but it's still worth it, since it fixes the
case we care about (the parser appending).
We could also optimize pure insertions (since in that case we can still figure
out what the old text was), but it's probably annoying and not worth the churn.
In any case, we cannot optimize anything that resembles any kind of removal,
because from there we don't know the old text in any way (and the text nodes
like to reuse string buffers and such).
We could do two other optimizations to replace / extend this one, in that order:
* Pass the buffer and length to CharacterDataWillChange, and use that to get
the exact old text and the new one in RestyleManager. That would make the
optimization exact.
* Pass some sort of Maybe<bool> mWasWhitespace down the CharacterDataChangeInfo
which is computed like:
HasFlag(NS_CACHED_TEXT_IS_ONLY_WHITESPACE)
? Some(NS_TEXT_IS_ONLY_WHITESPACE)
: Nothing()
It's not clear to me it's going to be completely worth the churn, so I haven't
done those yet, if we see code in the wild which resembles my testcase, we can
think of doing it.
MozReview-Commit-ID: 2rTWaZti8rv
--HG--
extra : rebase_source : 7390b8740801eb7b91700bb2533c43c173ac5db9
It is expected to use 64-bit for all the restyle generation counters, since the
getter methods all return uint64_t type at present. However, we're using uint32_t
for the actual counter variables, which means the potential overflow issue is not
avoided.
In this patch, we use 64-bit for the restyle generation counters, so the overflow
issue can be avoided as expected.
MozReview-Commit-ID: 2y2afIcuwvc
--HG--
extra : rebase_source : 3fe64d7d3fc00fa1031eef9f0c15b64405435dfd
(Path is actually r=froydnj.)
Bug 1400459 devirtualized nsIAtom so that it is no longer a subclass of
nsISupports. This means that nsAtom is now a better name for it than nsIAtom.
MozReview-Commit-ID: 91U22X2NydP
--HG--
rename : xpcom/ds/nsIAtom.h => xpcom/ds/nsAtom.h
extra : rebase_source : ac3e904a21b8b48e74534fff964f1623ee937c67
We should not be declaring forward declarations for nsString classes directly,
instead we should use nsStringFwd.h. This will make changing the underlying
types easier.
--HG--
extra : rebase_source : b2c7554e8632f078167ff2f609392e63a136c299
This patch changes UpdateAnimationOnlyStyles to only flush animation styles if
there are throttled animation styles that could affect hit-testing and renames
the function to UpdateAnimationStylesForHitTesting at the same time.
For GeckoRestyleManager, the original UpdateAnimationOnlyStyles which flushes
animation styles if there are any pending animation styles, is renamed to
UpdateAnimationStyles for consistency.
MozReview-Commit-ID: 89UleXjI2OE
--HG--
extra : rebase_source : cac59f9d48b096aee718f89ffe203d38d39027fd
Adds another restyle generation which represents the dirty state of raw
style changes, so that getComputedStyle() wont be confused by optimizations
made by style engines.
MozReview-Commit-ID: 7RYeNCzFygO
--HG--
extra : rebase_source : e2a6bd73f4b2b71115abe68b5cce95be57cbd9b2
