And fix a bunch of fallout as we now don't include nsDocShell.h everywhere.
Differential Revision: https://phabricator.services.mozilla.com/D62169
--HG--
extra : moz-landing-system : lando
Mostly what I was seeing come up in the .deps files from my object directory
which sounded feasible.
Differential Revision: https://phabricator.services.mozilla.com/D62167
--HG--
extra : moz-landing-system : lando
This avoids including nsFont.h and a bunch of other stuff in nsPresContext.h
Differential Revision: https://phabricator.services.mozilla.com/D62166
--HG--
extra : moz-landing-system : lando
We include it everywhere because it's included from gfxTypes.h.
This should avoid including all the generated bindings _everywhere_.
Differential Revision: https://phabricator.services.mozilla.com/D62174
--HG--
extra : moz-landing-system : lando
APZ may want to know what the main thread's view of the visual viewport offset
was at the time of the last paint even if the main thread does not want APZ
to scroll to that visual viewport offset.
Differential Revision: https://phabricator.services.mozilla.com/D62093
--HG--
extra : moz-landing-system : lando
`mIsCaretPositionChanged` is used in `UpdateShouldDisableApz()` to
determine whether to disable APZ during scrolling or pinch-zooming.
Suppose the selection is on position:static elements. When
pinch-zooming, the zoom level is changed, but the position is not. We
split `PositionChangedResult::Changed` into two separate states, and set
`mIsCaretPositionChanged` only when the position is changed.
Differential Revision: https://phabricator.services.mozilla.com/D61516
--HG--
extra : moz-landing-system : lando
This removes nsStyleImageRequest by moving the load state to LoadData instead
(where other lazy state like the resolved URL and load id lives).
That way we can use cbindgen for more stuff (there's no blocker for using it for
all images now), and we can undo the image tracking shenanigans that I had to do
in bug 1605803 in nsImageFrame.
This removes the mDocGroup member because well, there's no real upside of that
now that quantum DOM is not a thing.
It also removes the static clones of the image requests, and the need for each
computed value instance to have its own request. These were needed because we
needed the image loader for the particular document to observe the image
changes. But we were also tracking the request -> loader for other purposes.
Instead, Now all the images get loaded with GlobalImageObserver as a listener,
which looks in the image map and forwards the notification to all the interested
loaders instead dynamically.
The style value is only responsible to load the image, and no longer tracks /
locks it. Instead, the loader does so, via the image tracker.
Differential Revision: https://phabricator.services.mozilla.com/D58519
--HG--
extra : moz-landing-system : lando
To avoid computing transform bounds over and over. It is generally just better.
Replace the various "overridebounds" thingies by using the "fallback" of the
transform-reference-box code which we need anyway.
Differential Revision: https://phabricator.services.mozilla.com/D61890
--HG--
extra : moz-landing-system : lando
I thought this would fix <input type=number style="user-select: none">, but
turns out it doesn't.
<input type=number> doesn't have the editor root as a root of the anonymous
subtree, so the current hack wouldn't work, as the anon root wouldn't have the
editable flag. So tweak the code a bit to handle stuff in a simpler way than
setting the flags after the fact, and set the NAC-root flag earlier to avoid
the mOuterWrapper->AppendChildTo(root) call forgetting about root's flags.
I had to tweak one AccessibleCaret test, but that's because it uses <textarea>
with user-select: none, and our behavior there is not particularly sane. It just
happened to work because that test-case also had a bunch of contenteditable
elements, and we stop matching this rule:
https://searchfox.org/mozilla-central/rev/220a3bd6063fcbe5ca50e88dcabdc7dee0aca448/layout/style/contenteditable.css#22
Because the anonymous div now properly matches :-moz-read-write, which made the
rule apply and the test work. See comment 4 of this bug.
I'll fix this stuff up and add some tests for our behavior here in bug 1611699.
I refactored the dragdrop tests to cover more input types, but I ended up not
being able to use them because they're dependent on the content.
Instead I added an extra test and changed the refactor so that it applies to
<input type=search>, as there's layout work going on in bug 558594, and it'd be
unfortunate to regress this there too.
Differential Revision: https://phabricator.services.mozilla.com/D61094
--HG--
extra : moz-landing-system : lando
This removes nsStyleImageRequest by moving the load state to LoadData instead
(where other lazy state like the resolved URL and load id lives).
That way we can use cbindgen for more stuff (there's no blocker for using it for
all images now), and we can undo the image tracking shenanigans that I had to do
in bug 1605803 in nsImageFrame.
This removes the mDocGroup member because well, there's no real upside of that
now that quantum DOM is not a thing.
It also removes the static clones of the image requests, and the need for each
computed value instance to have its own request. These were needed because we
needed the image loader for the particular document to observe the image
changes. But we were also tracking the request -> loader for other purposes.
Instead, Now all the images get loaded with GlobalImageObserver as a listener,
which looks in the image map and forwards the notification to all the interested
loaders instead dynamically.
The style value is only responsible to load the image, and no longer tracks /
locks it. Instead, the loader does so, via the image tracker.
Differential Revision: https://phabricator.services.mozilla.com/D58519
--HG--
extra : moz-landing-system : lando
When our position changes from / to absolute / fixed, then we need to reframe
(most likely, at least). But in some easier cases we can just live with a reflow
plus (optionally) a containing-block update or a repaint, depending on the case.
We need to delete the normal position property when this happens and we switch
from static to relative/sticky, and also need to update the handling of
UpdateContainingBlock to avoid making decisions based on position not changing.
This will actually not cause more reframes, as one would expect, because that
optimization became somewhat obsolete by bug 1519371 (which made the
optimization exact).
Differential Revision: https://phabricator.services.mozilla.com/D61142
--HG--
extra : moz-landing-system : lando
Summary of the changes/reasons:
- LayoutTelemetryTools.h directly uses several types whose headers it needs to
include. (These includes were present in its .cpp file; I'm migrating them
from there to the .h file, and I'm adding a new include for "Saturate.h" to
provide the SaturateUint8 type.)
- LayoutTelemetryTools.cpp needs an include for MainThreadUtils.h, to provide
NS_IsMainThread().
- StaticPresData.cpp needs an include for ServoUtils.h, to provide
AssertIsMainThreadOrServoFontMetricsLocked().
- ZoomConstraintsClient.h needs a forward-decl for mozilla::dom::Document since
it uses a pointer of that type in a function-decl.
- ScrollSnap.h needs forward-decls of nsPoint/nsRect for some references to
those types in a method signature.
- nsGridContainerFrame.cpp needs an include for nsBoxLayoutState.h since it
uses that type (it instantiates a nsBoxLayoutState instance).
- nsPlaceholderFrame.cpp needs a "using" decl for the mozilla::dom namespace in
order to use the un-namespace-prefixed "Element" type.
Differential Revision: https://phabricator.services.mozilla.com/D61603
--HG--
extra : moz-landing-system : lando
This patch is generated via:
1. Manually modify gfx/2d/Types.h
2. Run the following script and clang-format.
3. Add brackets for the for loop in nsCSSRendering.cpp.
```
function rename() {
echo "Renaming $1 to $2"
rg -l "$1" | xargs sed -i -E -e s/"$1"/"$2"/g
}
rename "NS_FOR_CSS_HALF_CORNERS\(i\)" "for (const auto i : mozilla::AllPhysicalHalfCorners())"
rename "NS_FOR_CSS_HALF_CORNERS\(corner\)" "for (const auto corner : mozilla::AllPhysicalHalfCorners())"
```
Differential Revision: https://phabricator.services.mozilla.com/D61252
--HG--
extra : moz-landing-system : lando
`nsRange` instances are allocated a lot in the heap especially by editor and
spellchecker. The allocation cost is too bad for benchmarks. Therefore,
we should reuse released instances as far as possible. For managing it in
static factory methods of `nsRange`, we need to hide `nsRange` constructor.
Differential Revision: https://phabricator.services.mozilla.com/D61237
--HG--
extra : moz-landing-system : lando
LayoutFrameType::NumberControl is unused, and nsNumberControlFrame inherits now
from nsTextControlFrame. There are three places that check for
LayoutFrameType::TextInput. Two direct ones:
* ShouldApplyOverflowCLipping
* nsFontInflationData
And one indirect via IsTextInputFrame, the one mentioned in comment 2. For all
those three, it makes sense for nsNumberControlFrame to be handled in the same
way as nsTextControlFrame.
Long term we may want to get rid of the concept of frame types and just use
queryFrame or some sugar of that sort, as it's error prone. But for now this
fixes the bug.
Differential Revision: https://phabricator.services.mozilla.com/D61088
--HG--
extra : moz-landing-system : lando
It's possible that UpdateDynamicTollbarOffset is called repeatedly before we
actually fire a window resize event.
Differential Revision: https://phabricator.services.mozilla.com/D60812
--HG--
extra : moz-landing-system : lando
Per SVG2 spec, the EBNF allows the path data string to be empty.
An empty path data string disables rendering of the path.
Therefore, we should make path('') a valid path string.
The related spec issue: https://github.com/w3c/fxtf-drafts/issues/392.
Now we serialize `path("")` as `path("")` for offset-path and clip-path.
Differential Revision: https://phabricator.services.mozilla.com/D60771
--HG--
extra : moz-landing-system : lando
