These were rendering as blank with container scrolling, and their failure with
containerless scrolling is caused by an unrelated viewport bug, tracked in
bug 1527511.
Differential Revision: https://phabricator.services.mozilla.com/D19681
--HG--
extra : moz-landing-system : lando
Manual testing shows expected rendering of the pages in question. The tests
do not reproduce locally and are tricky to debug on Try. Moreover, most
other reftests in this directory are already marked as failing or random
on Android.
For these reasons, I don't want these failures to block the rollout of
containerless scrolling. Bug 1084564 tracks re-enabling them.
Differential Revision: https://phabricator.services.mozilla.com/D19679
--HG--
extra : moz-landing-system : lando
The Timestamp implementation on Windows seems to be based off of two system APIs: GetTickCount64 and QueryPerformanceCounter. QPC seems to be preferred over GTC, unless a "failure" in QPC is detected.
There is code in the Timestamp implementation to detect if QPC "fails" be comparing elapsed time of QPC to elapsed time of GTC.
Through logging on try server I observed the following in different failures:
1) the QPC failure detection kicks in a lot and disables use of QPC
2) the QPC of the vsync timestamp was in the future compared to the QPC of TimeStamp::Now().
3) the QPC of TimeStamp::Now() was missing (presumably because it was disabled) and the GTC of the vsync timestamp was in the future compared to the GTC of TimeStamp::Now()
4) the vsync timestamp printf of raw GTC and QPC values matched at the source (SoftwareDisplay::NotifyVsync) and in the refresh driver, so the vsync timestamp wasn't getting messed up.
I guess we could spend more time on the error detection in our Windows TimeStamp code, but this failure is only happening on 32-bit machines.
Here's a patch to disable the assertions on 32-bit Windows. Depends how important we think this assertion is relative to the work to investigate more.
Differential Revision: https://phabricator.services.mozilla.com/D22453
--HG--
extra : moz-landing-system : lando
The only fishy bit is the animation stuff. In particular, there are two places
where we just mint the revert behavior:
* When serializing web-animations keyframes (the custom properties stuff in
declaration_block.rs). That codepath is already not sound and I wanted to
get rid of it in bug 1501530, but what do I know.
* When getting an animation value from a property declaration. At that point
we no longer have the CSS rules that apply to the element to compute the
right revert value handy. It'd also use the wrong style anyway, I think,
given the way StyleBuilder::for_animation works.
We _could_ probably get them out of somewhere, but it seems like a whole lot
of code reinventing the wheel which is probably not useful, and that Blink
and WebKit just cannot implement either since they don't have a rule tree,
so it just doesn't seem worth the churn.
The custom properties code looks a bit different in order to minimize hash
lookups in the common case. FWIW, `revert` for custom properties doesn't seem
very useful either, but oh well.
Differential Revision: https://phabricator.services.mozilla.com/D21877
--HG--
extra : moz-landing-system : lando
`PresShell::EventHandler::HandleEventInternal()` needs to accumulate event
handling time per each event type. The handling start time needs to be
recorded before sending EventStateManager. Therefore, this patch makes the
helper class which is a stack class, records current time at construction
and calls `Telemetry::AccumulateTimeDelta()` at destruction automatically.
Differential Revision: https://phabricator.services.mozilla.com/D21335
--HG--
extra : moz-landing-system : lando
We always pass the same set of types so it seems simpler to just look it up
within the function itself.
Differential Revision: https://phabricator.services.mozilla.com/D22444
--HG--
extra : moz-landing-system : lando
If `aEvent` requires frame but there is no event target,
`PresShell::EventHandler::HandleEventInternal()` just records the response
time. So, we can reduce one indent level in the big method.
Note that I'm not sure recording the response time in such case because
the *good* values may make the average and median better. But this is
out of scope of bug 1466208.
Differential Revision: https://phabricator.services.mozilla.com/D21336
--HG--
extra : moz-landing-system : lando
This was introduced in bug 726817 for B2G stuff and there is no call site of
SetClampScrollPosition now and the only one entry point to call the function
was nsIFrameLoader.clampScrollPosition which was removed in
https://hg.mozilla.org/mozilla-central/rev/cd9aace7bb96 .
ScrollFrameHelper::ShouldClampScrollPosition() had always returned true in both
cases of root and non-root scrollable frames. (In the case of non-root
scrollable, nsFrameLoader.mClampScrollPosition is initially set to true and
nobody changes the value)
Differential Revision: https://phabricator.services.mozilla.com/D22026
--HG--
extra : moz-landing-system : lando
`PresShell::EventHandler::HandleEventInternal()` recodes event handling
response performance with telemetry after it dispatches the event. We can move
it into new method simply.
Differential Revision: https://phabricator.services.mozilla.com/D21334
--HG--
extra : moz-landing-system : lando
The existing linear gradient shader is quite slow, especially
on very large gradients on integrated GPUs.
The vast majority of gradients in real content are very simple
(typically < 4 stops, no angle, no repeat). For these, we can
run a fast path to persist a small gradient in the texture cache
and draw the gradient via the image shader.
This is _much_ faster than the catch-all gradient shader, and also
results in better batching while drawing the main scene.
In future, we can expand the fast path to handle more cases, such
as angled gradients. For now, it takes a conservative approach,
but still seems to hit the fast path on most real content.
Differential Revision: https://phabricator.services.mozilla.com/D22445
--HG--
extra : moz-landing-system : lando
Calling getComputedStyle on an element outside of the document always returns
the empty string since bug 1511138, so we were incorrectly thinking that the
initial font-family was serif when it was not.
I chose a <meta> element instead of a <div> because they're undisplayed and they
don't cause the <body> of the document to start (though I _think_ that only
happens from the parser).
In any case I'm pretty surprised this didn't cause issues in other tests...
Differential Revision: https://phabricator.services.mozilla.com/D22604
--HG--
extra : moz-landing-system : lando
If `Shift` state of `eContextMenu` event is active, we make it not fired on
web content. Additionally, if it's not time to open context menu, we shouldn't
dispatch it into the DOM. The new method prepare and check them.
Differential Revision: https://phabricator.services.mozilla.com/D21337
--HG--
extra : moz-landing-system : lando
We can get back the fancy flag syntax as soon as we get C++17 inline variables,
which I sent an email to dev-platform@ about, with no reply.
Differential Revision: https://phabricator.services.mozilla.com/D22382
--HG--
extra : moz-landing-system : lando
This maintains the important invariant that layers that carry scroll metadata
for the RCD-RSF are inside the async zoom container.
Differential Revision: https://phabricator.services.mozilla.com/D22049
--HG--
extra : moz-landing-system : lando
We really only have two sets of prefs, one for chrome-like documents
(stuff in chrome docshells + chrome-origin images), and one for the rest.
Differential Revision: https://phabricator.services.mozilla.com/D20946
--HG--
extra : moz-landing-system : lando
"browser.firstrun.*" seems to have been unused since the end of XUL-based
Fennec, whereas the code referencing the "browser.snippets.*" prefs was removed
in bug 1482836.
Differential Revision: https://phabricator.services.mozilla.com/D20862
--HG--
extra : moz-landing-system : lando
We can totally get there with a negative percentage basis, see comment 0 for an
example.
We could keep the warning like:
NS_WARNING_ASSERTION(clamping_mode == StyleAllowedNumericType::All ||
basis >= 0, "nscoord overflow?");
Which will catch cases where the style system would refuse to parse a negative
<length-percentage>, but we got a negative percent basis, which would be weird.
But that's a bit misleading since right now at least we rely on the caller to do
the appropriate clamping. I also think that NS_WARNING_ASSERTION is not very
useful, since we're not very likely to catch stuff with it. But anyhow, your
call.
Differential Revision: https://phabricator.services.mozilla.com/D22328
--HG--
extra : moz-landing-system : lando
The remaining part of `PresShell::EventHandler::HandleEvent()` does:
1. Handles the event at focused content.
2. Handles the event with given frame which is a frame for `mPresShell`.
For making them clearer, this patch moves them into new methods.
Differential Revision: https://phabricator.services.mozilla.com/D21197
--HG--
extra : moz-landing-system : lando
There aren't any manual changes here; this is just from me running:
./mach clang-format -p layout/style/nsStyleConsts.h
Differential Revision: https://phabricator.services.mozilla.com/D22211
--HG--
extra : moz-landing-system : lando
If focused element is in another document,
`PresShell::EventHandler::HandleEvent()` needs to retarget the event to another
`PresShell`. This patch moves the case into new overload method,
`MaybeHandleEventWithAnotherPresShell()`.
Additionally, removes `PresShell::HandleRetargetedEvent()` and makes
`EventHandler::HandleRetargetedEvent()` non-public because the new method
is the only user of them.
Differential Revision: https://phabricator.services.mozilla.com/D21196
--HG--
extra : moz-landing-system : lando
This lets us restrict the size of the mask surface that we allocate
to the destination tile size. This gives a large performance
improvement.
It also includes some miscelanous fixes to the CreateClippedDrawTarget
code path.
Differential Revision: https://phabricator.services.mozilla.com/D21750
DTO = DrawTargetOffset = (-Origin.x, -Origin.y)
CTM = CurrentTransform
MASKT = MaskTransform = Inv(CurrentTransform)
MASKO = MaskOffset = (GetRect().x, GetRect().y)
The transform to device space is:
MASKT * CTM * DTO
Currently we do:
MASKT = MASKT * MASKO
Which gives us MASKT * MASKO * CTM * DTO. This works fine if there's no scale but
will not properly cancel out if there is.
I believe we want:
MASKO * MASKT * CTM * DTO
To get this we need to preTranslate instead of postTranslate.
Differential Revision: https://phabricator.services.mozilla.com/D21604
Since bug 1524480 we set the NS_FRAME_MAY_BE_TRANSFORMED frame bit when needed
in RestyleManager::ProcessRestyledFrames so that it is now redundant to also set
it from KeyframeEffect.
Furthermore, setting frame bits from KeyframeEffect is a little fragile since it
depends on the life cycle of the KeyframeEffect which is independent of the
nsFrame. If we can avoid doing that, we probably should.
Differential Revision: https://phabricator.services.mozilla.com/D21885
--HG--
extra : moz-landing-system : lando
As with the previous patch in this series, we need to pay particular attention
to how we handle display:table content when detecting animations on a element.
Please see the extended description in that patch for an explanation of
different frame types involved.
As with transforms, our handling of opacity is also inconsistent. In
particular, we fail to return true from nsIFrame::HasOpacityInternal for
display:table content with opacity animations applied due to the conflicting
requirements for a primary frame and having opacity animations (which are stored
on the style frame).
Unlike transforms, however, we do not inherit the opacity to the table wrapper.
Instead we leave it on the inner table frame. As a result, we should not check
for a primary frame, but instead we should check for the style frame for the
primary frame.
This patch adjusts this handling to check instead for the appropriate style
frame as opposed to requiring a primary frame. It includes a reftest that fails
without the code changes in this patch.
Differential Revision: https://phabricator.services.mozilla.com/D21884
--HG--
extra : moz-landing-system : lando
For display:table content we generate two frames: a table wrapper frame and an
inner table frame. The styles are applied to the inner frame (referred to as the
style frame), whilst the wrapper frame is the primary frame for the content.
However, in order to make tables with transforms behave as a container for
abspos/fixed-pos content as required by the spec, we apply the transform to the
wrapper frame (bug 722777) by inheriting the transform from inner to wrapper and
then ignoring the transform on the inner frame (bug 722777 and bug 816458).
When handling animations on table elements we need to be careful of this
distinction. in particular, css animations[1] and web animations[2] require that
when we have an unfinished transform animation targetting an element, the
element acts as if it had `will-change: transform` applied and therefore
generates a stacking context. As a result we need to accurately detect when
a frame should be considered as having transform animations applied to it or not
for the purpose of creating a stacking context.
Previously our handling of display:table content was quite inconsistent and
contradictory. For example, `nsIFrame::HasAnimationOfTransform` would check for
a primary frame AND for animations on that frame, despite the fact that we only
ever store animations on the style frame. As a result it could never return true
for either a table wrapper or inner table frame.
This patch attempts to make this handling at least a little more consistent,
producing the following result:
Outer table frame (primary frame):
nsIFrame::IsTransformed → true
nsIFrame::IsCSSTransformed → true
nsIFrame::HasAnimationOfTransform → true
nsLayoutUtils::HasAnimationOfProperty(frame, eCSSProperty_transform) → false
Inner table frame (style frame):
nsIFrame::IsTransformed → false
nsIFrame::IsCSSTransformed → false
nsIFrame::HasAnimationOfTransform → false
nsLayoutUtils::HasAnimationOfProperty(frame, eCSSProperty_transform) → true
We maintain that the NS_FRAME_MAY_BE_TRANSFORMED bit is only set on the primary
frame whilst the mMayHaveTransformAnimation flag is only set on the style frame.
Note that we don't simply always put everything on the primary frame because for
other property types (e.g. opacity) the default setup of putting all styles and
animations on the style frame is simpler and correct. So far it is only
transforms that require special handling to apply the effect to the wrapper
frame.
This patch adds a reftest that fails without the code changes included in this
patch.
[1] https://drafts.csswg.org/css-animations/#animations
[2] https://drafts.csswg.org/web-animations-1/#side-effects-section
Differential Revision: https://phabricator.services.mozilla.com/D21883
--HG--
extra : moz-landing-system : lando
With splitting `HandleEvent()` a lot, it becomes more difficult to keep
managing each set of calling `PushCurrentEventInfo()` and
`PopCurrentEventInfo()`. So, `EventHandler` should have a helper class
to push and pop current event info into/from the stack.
Differential Revision: https://phabricator.services.mozilla.com/D21198
--HG--
extra : moz-landing-system : lando
The calculation of aOverflow rect for a placeholder contained in a transformed
stacking context didn't take the transform into account, leading to an incorrect
rect which artificially inflated the calculated rebuild region.
Differential Revision: https://phabricator.services.mozilla.com/D21168
--HG--
extra : moz-landing-system : lando
This patch avoids a potential division by 0 (one that's unlikely to be
triggered by real content), for correctness and robustness.
This patch isn't really changing the logic, because the newly-guarded code is
already guarded by a "length < sum" check, and "length" is expected to be
nonnegative [1], which means "sum" would already have been nonzero in cases
where the existing strict less-than comparison returned true.
[1] One special case where we can have negative `length` values here:
if this is an outline with an `outline-offset` that is sufficiently negative
to entirely cancel out the outlined element's border-box size, then that can
produce a negative `length` (i.e. the outline box will have a negative size,
conceptually). In that scenario, we were previously dividing by 0 here and
working with bogus huge/infinite negative values as a result, but we should
not do that.
Depends on D21578
Differential Revision: https://phabricator.services.mozilla.com/D21579
--HG--
extra : moz-landing-system : lando
This patch doesn't change any logic/behavior -- it's just adding braces for clarity.
Differential Revision: https://phabricator.services.mozilla.com/D21578
--HG--
extra : moz-landing-system : lando
The way we control APZ zooming in WebRender is by inserting an animation
property placeholder on a WR stacking context, and then having APZ
update the animation transform value with the proper matrix at composite
time.
Previously, the stacking context being used was the rootmost
stacking context in the content process. However this doesn't work for
zoomable content in the UI process (e.g. about:support), and after
recent changes, also wraps display items that should not be affected by
zoom (e.g. scrollbars or the background color item).
This patch moves the animation property placeholder so that it
corresponds to the newly added nsDisplayAsyncZoom display item, which
corrects both of the above problems and is conceptually in line with the
desired behaviour.
Differential Revision: https://phabricator.services.mozilla.com/D21795
--HG--
extra : moz-landing-system : lando
This also adopts the resolution from [1] while at it, making letter-spacing
compute to a length, serializing 0 to normal rather than keeping normal in the
computed value, which matches every other engine.
This removes the SMIL tests for percentages from letter-spacing since
letter-spacing does in fact not support percentages, so they were passing just
by chance.
[1]: https://github.com/w3c/csswg-drafts/issues/1484
Differential Revision: https://phabricator.services.mozilla.com/D21850
--HG--
extra : moz-landing-system : lando
The existing ANR and tombstone checks in remoteautomation are used
by mochitests and reftests but are awkward for other harnesses like cppunit
that do not use remoteautomation...and easily missed. This patch moves that
code to the mozharness AndroidMixin, treating ANRs and tombstones like the
logcat: Make sure any old logs are deleted when the mozharness script starts,
then move any logs found at the end of the run to the upload directory.
https://treeherder.mozilla.org/#/jobs?repo=try&tier=1%2C2%2C3&revision=f3de1e9836da2b0c9232f5d92c751b979459e19b
demonstrates tombstone artifacts for the Android 7.0 cppunit tests.
Differential Revision: https://phabricator.services.mozilla.com/D21774
--HG--
extra : moz-landing-system : lando
clang's -Wmissing-prototypes option identifies global functions that can be made static (because they're only called from one compilation unit) or removed (if they're never called).
layout/build/nsLayoutModule.cpp:95:6 [-Wmissing-prototypes] no previous prototype for function 'nsLayoutModuleInitialize'
layout/build/nsLayoutModule.cpp:176:11 [-Wmissing-prototypes] no previous prototype for function 'CreateXMLContentSerializer'
layout/build/nsLayoutModule.cpp:178:11 [-Wmissing-prototypes] no previous prototype for function 'CreateHTMLContentSerializer'
layout/build/nsLayoutModule.cpp:180:11 [-Wmissing-prototypes] no previous prototype for function 'CreateXHTMLContentSerializer'
layout/build/nsLayoutModule.cpp:182:11 [-Wmissing-prototypes] no previous prototype for function 'CreatePlainTextSerializer'
layout/build/nsLayoutModule.cpp:184:11 [-Wmissing-prototypes] no previous prototype for function 'CreateContentPolicy'
layout/build/nsLayoutModule.cpp:188:11 [-Wmissing-prototypes] no previous prototype for function 'CreateGlobalMessageManager'
layout/build/nsLayoutModule.cpp:189:11 [-Wmissing-prototypes] no previous prototype for function 'CreateParentMessageManager'
layout/build/nsLayoutModule.cpp:191:11 [-Wmissing-prototypes] no previous prototype for function 'CreateChildMessageManager'
layout/build/nsLayoutModule.cpp:225:10 [-Wmissing-prototypes] no previous prototype for function 'Construct_nsIScriptSecurityManager'
layout/build/nsLayoutModule.cpp:237:10 [-Wmissing-prototypes] no previous prototype for function 'LocalStorageManagerConstructor'
layout/build/nsLayoutModule.cpp:255:6 [-Wmissing-prototypes] no previous prototype for function 'nsLayoutModuleDtor'
Differential Revision: https://phabricator.services.mozilla.com/D21854
--HG--
extra : rebase_source : 000c6f3cc49acb942a84af13037f7854f14af5e4
extra : histedit_source : 997598f1d82099cc195fc1bff23ffad9ffff7fdb
clang's -Wmissing-prototypes option identifies global functions that can be made static (because they're only called from one compilation unit) or removed (if they're never called).
NS_NewLayoutDebugger() is defined in nsLayoutDebugger.cpp.
layout/base/nsLayoutDebugger.cpp:44:10 [-Wmissing-prototypes] no previous prototype for function 'NS_NewLayoutDebugger'
Differential Revision: https://phabricator.services.mozilla.com/D21853
--HG--
extra : rebase_source : d56bdbce2030fdd3d036493c32af69addc6d1b73
extra : histedit_source : 61343f17eb6d5a774c848b1694e6507b7c1239ed
clang's -Wmissing-prototypes option identifies global functions that can be made static (because they're only called from one compilation unit) or removed (if they're never called).
The Gecko_*CrashReport() function definitions were added in bug 1407080. The function calls were removed in https://hg.mozilla.org/integration/autoland/rev/22f60f0b91ee. A clue that this function was no longer called: bug 1348273 changed the function prototype in GeckoBindings.h to take a uint32_t key parameter, but left the function definition in GeckoBindings.cpp taking a const char* key. And yet nothing blew up.
layout/style/GeckoBindings.cpp:2294:6 [-Wmissing-prototypes] no previous prototype for function 'Gecko_AnnotateCrashReport'
Differential Revision: https://phabricator.services.mozilla.com/D21852
--HG--
extra : rebase_source : 1ca8100716900f4b39504b743f4b06a17f6be832
extra : histedit_source : d662b34fdc783abe95d56ea1d6033ccc34d2ae38
clang's -Wmissing-prototypes option identifies global functions that can be made static (because they're only called from one compilation unit) or removed (if they're never called).
layout/painting/FrameLayerBuilder.cpp:2593:6 [-Wmissing-prototypes] no previous prototype for function 'ReleaseLayerUserData'
layout/painting/RetainedDisplayListBuilder.cpp:1084:6 [-Wmissing-prototypes] no previous prototype for function 'FindContainingBlocks'
layout/style/FontFaceSet.cpp:423:6 [-Wmissing-prototypes] no previous prototype for function 'IsPdfJs'
layout/svg/SVGObserverUtils.cpp:969:6 [-Wmissing-prototypes] no previous prototype for function 'DestroyFilterProperty'
Differential Revision: https://phabricator.services.mozilla.com/D21851
--HG--
extra : rebase_source : 7b876a976087eef710ff70ca0ae42abb3c505e49
extra : histedit_source : 4254fd49e36bef38bbb6197861a4a22e040d17e0
They were revised in bug 1158392 to work with containerless scrolling and
not with container scrolling.
Differential Revision: https://phabricator.services.mozilla.com/D22744
--HG--
extra : moz-landing-system : lando
They are only failing with containerless scrolling, which is nightly-only
for now.
Differential Revision: https://phabricator.services.mozilla.com/D22732
--HG--
extra : moz-landing-system : lando
Most remaining code in `PresShell::EventHandler::HandleEvent()` is what computes
event target of the event which should be handled on focused content. This
patch moves the part to the new method.
Additionally, moves `nsIPresShell::gKeyDownTarget` to
`EventHandler::sLastKeyDownEventTargetElement` and make it use `StaticRefPtr`.
Finally, for using `Element*` instead of `nsIContent*`, changes the result type
of `Document::GetUnfocusedKeyEventTarget()` to `Element*`.
Differential Revision: https://phabricator.services.mozilla.com/D21195
--HG--
extra : moz-landing-system : lando
Although debug build of Android/arm runs on emulator, test_parse_rule.html
sometimes runs over 5 min. So we should extend timeout value for this.
Differential Revision: https://phabricator.services.mozilla.com/D21663
--HG--
extra : moz-landing-system : lando
nsIconChannel (for moz-icon:// images) is unsound, see bug 1438939.
nsMenuPopupFrame::Init is also unsound on mac, looks like...
I'll try to get them fixed on trunk, but it's not worth crashing release for
this IMO, given it's pre-existing. The assert in PresShell::~PresShell hopefully
avoids exploitable issues.
Differential Revision: https://phabricator.services.mozilla.com/D20945
--HG--
extra : moz-landing-system : lando
urlExtraReferrer could be null, we should check before dereference. Otherwise
it will cause crash when webrender is enabled by gfx.webrender.enabled=true
and gfx.webrender.all=true,
Differential Revision: https://phabricator.services.mozilla.com/D20805
--HG--
extra : moz-landing-system : lando
When the event is not handled with coordinates and there is no frame for
`mPresShell`, `PresShell::EventHandler::HandleEvent()` handles the events
simpler than the case there is a frame. Therefore, this patch moves the
`else` block of `if (aFrame)` and reduce the indent of `if (aFrame)` case.
Differential Revision: https://phabricator.services.mozilla.com/D21194
--HG--
extra : moz-landing-system : lando
As it turns out we need this to avoid losing precision both during painting and
during serialization.
This patch also changes to serialize `context-value` if it's the computed value.
I could keep the previous behavior, but it makes no sense to serialize the
initial value. We're the only ones to support this value anyway, and I couldn't
find a definition or spec for this.
Also update tests and expectations for:
* New unexpected passes.
* Always serializing the unit in getComputedStyle.
* Calc and interpolation support.
Chrome also always serializes the unit in getComputedStyle, so I'm pretty sure
this is compatible with them. Chrome is inconsistent and keeps numbers in
specified style, but that's inconsistent with itself and with other quirky
lengths, so I updated the tests instead.
Differential Revision: https://phabricator.services.mozilla.com/D21819
Instead of storing them as LengthPercentage | Number, always store as
LengthPercentage, and use the unitless length quirk to parse numbers instead.
Further cleanups to use the rust representation can happen as a followup, which
will also get rid of the boolean argument (since we can poke at the rust length
itself). That's why I didn't bother to convert it to an enum class yet.
Differential Revision: https://phabricator.services.mozilla.com/D21804
Now, the block in HandleEvent(), which handles event using coordinates is
less than 200 lines. Perhaps, this is good amount to be split to a method.
This patch just moves the block to a new method.
Differential Revision: https://phabricator.services.mozilla.com/D21193
--HG--
extra : moz-landing-system : lando
After dispatching pointer events, `PresShell::EventHandler::HandleEvent()`
updates event target only when the event is a touch event. We should do it in
a new method of `EventTargetData`.
Although I don't know why this is done in
`PresShell::EventHandler::DispatchPrecedingPointerEvent()`.
Differential Revision: https://phabricator.services.mozilla.com/D21192
--HG--
extra : moz-landing-system : lando
We restrict the argument (i.e. `nsCSSPropertyIDSet`) of some functions is
the subset of other property set (e.g. transform-like properties), so
add this function for better readability.
Depends on D19629
Differential Revision: https://phabricator.services.mozilla.com/D21598
--HG--
extra : moz-landing-system : lando
Transform display item may have multiple properties, so it's better to
use display item type as the input.
Also, factor some code out of AddAnimationsForProperty, so we can easier
to extend this for multiple properties.
We will pass a list of layers::Animation to the compositor thread. In
this list, the animations belonging to the same property are grouped together,
so we can easily separate the animations by property and sample the animations
for each property on the compositor thread. (Will do this in Bug 1425837.)
Depends on D19628
Differential Revision: https://phabricator.services.mozilla.com/D19629
--HG--
extra : moz-landing-system : lando
We use DisplayItemType as the input of HasAnimationsForCompositor, and
nsCSSPropertyIDSet as the input of GetAnimationsForCompositor.
The caller of HasAnimationsForCompositor just wants to check if there is
any compositor animation for a display item, so we can replace it by the
display item, and get the properties from this display item.
However, the caller of GetAnimationsForCompositor may use a subset of
transform-like properties for getting scale factors, or use all the
transform-like properties for sending all transform animations to the
compositor thread.
Depends on D19630
Differential Revision: https://phabricator.services.mozilla.com/D19628
--HG--
extra : moz-landing-system : lando
FrameLayerBuilder needs to clear this flag by DisplayItemType, so we add
a new function for it.
Differential Revision: https://phabricator.services.mozilla.com/D19630
--HG--
extra : moz-landing-system : lando
Now, we can move the block dispatching preceding pointer event to separated
method. Then, we can hide the complicated retarget process after dispatching
a pointer event from HandleEvent().
Differential Revision: https://phabricator.services.mozilla.com/D21191
--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
--HG--
extra : moz-landing-system : lando
Currently, PresShell::EventHandler::HandleEvent() sets `overrideClickTarget`
only when Pointer Events is enabled and there is pointer capturing content,
and this is computed while dispatching a pointer event.
So, if we move it into EventTargetData, we can move the pointer event
dispatching block into a separated method and caller can receive it with
an EventTargetData instance which is anyway necessary to receive new
target frame after dispatching a pointer event.
Differential Revision: https://phabricator.services.mozilla.com/D21190
--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 patch avoids a potential division by 0 (one that's unlikely to be
triggered by real content), for correctness and robustness.
This patch isn't really changing the logic, because the newly-guarded code is
already guarded by a "length < sum" check, and "length" is expected to be
nonnegative [1], which means "sum" would already have been nonzero in cases
where the existing strict less-than comparison returned true.
[1] except for integer overflow or other bizarreness.
Depends on D21578
Differential Revision: https://phabricator.services.mozilla.com/D21579
--HG--
extra : moz-landing-system : lando
This patch doesn't change any logic/behavior -- it's just adding braces for clarity.
Differential Revision: https://phabricator.services.mozilla.com/D21578
--HG--
extra : moz-landing-system : lando
We cannot move each block into separated methods while computing EventTargetData
because we need to check capturing contents, etc. Therefore, only each block
should be moved to separated methods for now.
This moves a block which computes event target from point of the event. If
this can be moved to EventTargetData, it might be easier to understand, but
its helper method GetFrameToHandleNonTouchEvent() requires to access members
of EventHandler. Therefore, we need to treat EventTargetData as an out param
of the new method.
Differential Revision: https://phabricator.services.mozilla.com/D21189
--HG--
extra : moz-landing-system : lando
The main behavior changes are:
1) We no longer create a new Window when doing document.open(). We use the
same Window but remove all the event listeners on it and on the existing DOM
tree before removing the document's existing kids.
2) We no longer create a new session history entry. The existing one always
gets replaced instead.
3) We now support document.open on documents that are not in a Window.
The reasons for the various test changes are as follows:
The change to browser_modifiedclick_inherit_principal.js is because we no
longer set the docshell to a wyciwyg URL when document.open() happens and the
test was depending on that to terminate.
browser_wyciwyg_urlbarCopying.js is being removed because it's trying to test
wyciwyg URIs, which no longer exist.
The changes in docshell/test/navigation are because document.open() no longer
affects session history. One of the tests was testing the interactions there
and is being removed; another is being repurposed to just test that
document.open() does not affect history.length.
The change to test_x-frame-options.html is because document.open() now removes
event listeners on the window, which it didn't use to do (and in the specific
case in this test reused the existing inner too, so the listener was still
around in practice). The new behavior matches other browsers.
The removal of test_bug172261.html is because document.open() no longer affects
session history, so you can't go back across it or forward to the "opened"
state, so the situation that test is trying to test no longer exists.
The changes to test_bug255820.html are because reloading a document after
document.open() will now just load the URL of the document that was the entry
document for the open() call, not reload the written content. So there's not
much point testing reload behavior, and in this test it was just reloading the
toplevel test file inside the frames.
The change to test_bug346659.html is because now we no longer create a new
Window on document.open().
The change to test_bug1232829.html is because document.open() (implicit in this
test) no longer adds history entries, so the back() was just leaving the test
page instead of going back across the document.open(). The test is a
crashtest in practice, so might still be testing something useful about how
document.open() interacts with animations.
The change to test_bug715739.html is because the URL of the document after
document.open() is now the URL of the entry document, not a wyciwyg URL, so
reload() has different behavior than it used to.
The change to test_bug329869.html is because now when we go back we're
reloading the original document we had, not doing a wyciwyg load, and the
security info now doesn't include the untrusted script.
The changes to the wpt expectations are removing a bunch of expected failures
now that we pass those tests and disabling some tests that are fundamentally
racy and hence fail randomly. The latter all have github issues filed for the
test problem.
The change to testing/web-platform/tests/common/object-association.js is fixing
tests that were not matching the spec (and were failing in other browsers).
The change to parser-uses-registry-of-owner-document.html is fixing tests that
were not matching the spec (and were failing in other browsers).
The change to document-write.tentative.html is because the test was buggy: it
was using the same iframe element for all its tests and racing loads from some
tests against API calls from other tests, etc. It's a wonder it ever managed
to pass, independent of these patches (and in fact it doesn't pass according to
wpt.fyi data, even in Firefox).
The changes in html/browsers/history/the-history-interface are because
document.open() no longer adds history entries. The test was failing in all
other browsers for the same reason.
The changes in html/browsers/history/the-location-interface are because
reloading a document.open()-created thing now loads the URL of the page that
was the entry document for the open() call. The test was failing in all other
browsers.
The change to reload_document_open_write.html is because we now reload the url
of the document that entered the script that called open() when we reload, not
the written content. Other browsers were failing this test too; Gecko with
the old document.open implementation was the only one that passed.
The change to http-refresh.py is to fix a test bug: it was not returning a
Content-Type header, so we were putting up helper app dialogs, etc.
The change to test_ext_contentscript.js is because we no create a new global
for document.open() calls. Kris Maglione OKed this part.
Differential Revision: https://phabricator.services.mozilla.com/D17323
--HG--
extra : moz-landing-system : lando
As per the following change to the HTML spec:
86b05f8a07
when running a requestAnimationFrame callback it should be possible to cancel
another requestAnimationFrame callback scheduled to run in the same frame by
using cancelAnimationFrame.
See issue:
https://github.com/whatwg/html/issues/4359
Differential Revision: https://phabricator.services.mozilla.com/D20974
--HG--
extra : moz-landing-system : lando
In the next patch in this series we want to compare the handle of frame
callbacks we are about to run, with a set of canceled handles stored on the
document. This patch makes us pass the handles along with the callbacks so we
can do that.
Incidentally doing this allows us to just swap array elements when building up
the refresh driver's set of callbacks to run. That is hopefully a little more
efficient than running the implicit conversion operator on each item and then
appending to an array.
Differential Revision: https://phabricator.services.mozilla.com/D20973
--HG--
extra : moz-landing-system : lando
As per the following change to the HTML spec:
86b05f8a07
when running a requestAnimationFrame callback it should be possible to cancel
another requestAnimationFrame callback scheduled to run in the same frame by
using cancelAnimationFrame.
See issue:
https://github.com/whatwg/html/issues/4359
Differential Revision: https://phabricator.services.mozilla.com/D20974
--HG--
extra : rebase_source : ff3c69a82f0ef61f562edd1610017a17c8f26276
In the next patch in this series we want to compare the handle of frame
callbacks we are about to run, with a set of canceled handles stored on the
document. This patch makes us pass the handles along with the callbacks so we
can do that.
Incidentally doing this allows us to just swap array elements when building up
the refresh driver's set of callbacks to run. That is hopefully a little more
efficient than running the implicit conversion operator on each item and then
appending to an array.
Differential Revision: https://phabricator.services.mozilla.com/D20973
--HG--
extra : rebase_source : 4786ffedf33051298c9593a588a6e8ce70cabf04
And while we're at it, update the reference case to have "Reference" in
its title, so that the reference case and testcases are easier to distinguish
when viewing them side-by-side in several tabs.
Differential Revision: https://phabricator.services.mozilla.com/D21045
--HG--
extra : moz-landing-system : lando
Apparently we sometimes fall back to skia canvas with WebRender if ANGLE
fails to initialize. This makes the azureSkia condition true, but the
test still passes because of WebRender's magical properties. This patch
updates the annotation to reflect reality, so we don't get unexpected
passes.
Differential Revision: https://phabricator.services.mozilla.com/D21012
--HG--
extra : moz-landing-system : lando
Previously, if the accesskey attribute was missing then the label would reach up
to binding parent to find it's accesskey. In practice, bindings already do
[xbl:inherits=accesskey] to send it down to the label anyway.
The problem with this is that for controls without accesskeys, the attribute doesn't get set,
so the label will access the control from JS. This is fine for XBL, since typically
the label XBL will construct at the same time as the control, but when migrating
to Custom Elements, the label gets connected even when the control is hidden.
Differential Revision: https://phabricator.services.mozilla.com/D12355
--HG--
extra : moz-landing-system : lando
It's easier to see what remains that way. Done with the following script:
```
execfile("layout/style/ServoCSSPropList.py")
for p in data:
if p.type() != "longhand":
continue
if "GetCSNeedsLayoutFlush" in p.flags or "SerializedByServo" in p.flags or "Internal" in p.flags:
continue
print(p.name)
```
Ran like:
```
$ python print.py | sort
```
From the objdir.
Differential Revision: https://phabricator.services.mozilla.com/D20965
It's easier to see what remains that way. Done with the following script:
```
execfile("layout/style/ServoCSSPropList.py")
for p in data:
if p.type() != "longhand":
continue
if "GetCSNeedsLayoutFlush" in p.flags or "SerializedByServo" in p.flags or "Internal" in p.flags:
continue
print(p.name)
```
Ran like:
```
$ python print.py | sort
```
From the objdir.
Differential Revision: https://phabricator.services.mozilla.com/D20965
These casts are now unnecessary after Part 2, due to all consumers directly
using `ContentParent`.
Depends on D20551
Differential Revision: https://phabricator.services.mozilla.com/D20552
--HG--
extra : moz-landing-system : lando
Apparently we sometimes fall back to skia canvas with WebRender if ANGLE
fails to initialize. This makes the azureSkia condition true, but the
test still passes because of WebRender's magical properties. This patch
updates the annotation to reflect reality, so we don't get unexpected
passes.
Differential Revision: https://phabricator.services.mozilla.com/D21012
--HG--
extra : moz-landing-system : lando
These casts are now unnecessary after Part 2, due to all consumers directly
using `ContentParent`.
Depends on D20551
Differential Revision: https://phabricator.services.mozilla.com/D20552
--HG--
extra : moz-landing-system : lando
This is a new implementation of mix-blend compositing that is meant to be more idiomatic to WR and efficient.
Previously, mix-blend mode was composed in the following way:
1. parent stacking context was forced to isolate
2. source picture is also isolated
3. when rendering the isolated context, the framebuffer is read upon reaching the source. Both the readback and the source are placed in the RT cache.
4. a mix-blend draw call is issued to read from those cache segments and blend on top of the backdrop
The new implementation works by using the picture cutting (intruduced for preserve-3D contexts earlier) and some bits of magic:
1. backdrop stacking context is isolated with a special composition mode that prevents it from actually rendeing unless the suorce stacking context is invisible.
2. source stacking context is isolated with mix-blend composition mode that has a pointer to the backdrop picture
3. the instance of the backdrop picture is placed as a peer of the source picture (not a child)
4. if the backdrop is invisible, the source is drawn as a simple blit
5. otherwise, it's a draw call that reads from the isolated backdrop and source textures
Note the differences:
- parent stacking context is not isolated, but backdrop is
- no framebuffer readback is involved
- the source and backdrop pictures are rendered in parallel in a pass, improving the batching
- we don't blend onto the backdrop while reading from the backdrop copy at the same time
- the depth of the render pass tree is reduced: previously the parent and the source were isolated, now the source and the backdrop, which are siblings
Differential Revision: https://phabricator.services.mozilla.com/D20608
--HG--
rename : gfx/wr/wrench/reftests/blend/multiply-2-ref.yaml => gfx/wr/wrench/reftests/blend/multiply-3-ref.yaml
rename : gfx/wr/wrench/reftests/blend/multiply-3.yaml => gfx/wr/wrench/reftests/blend/multiply-4.yaml
extra : moz-landing-system : lando
All transform-like properties should set PositioningPropertyMutated.
Depends on D19632
Differential Revision: https://phabricator.services.mozilla.com/D19635
--HG--
extra : moz-landing-system : lando
Add comments and property check for the caller of ActivityTracker.
Depends on D19525
Differential Revision: https://phabricator.services.mozilla.com/D19632
--HG--
extra : moz-landing-system : lando
So we can let KeyframeEffect::ContainsAnimatedScale check individual
transforms, which is used by ActiveLayerTracker.
Depends on D19631
Differential Revision: https://phabricator.services.mozilla.com/D19525
--HG--
extra : moz-landing-system : lando
Let ActiveLayerTracker track individual transforms. (Will add
motion-path in the future.)
Besides, using a property set for transform and opacity is more efficient,
so let's change it. For background position, we use a different code path,
so we can have more restrictions in IsStyleAnimated.
Differential Revision: https://phabricator.services.mozilla.com/D19631
--HG--
extra : moz-landing-system : lando
If there's ::first-letter pseudo element on the multicol container,
ColumnSetWrapperFrame will have the first letter style, but it won't get
the NS_BLOCK_HAS_FIRST_LETTER_STYLE bit during frame construction. The
actual first-letter frame construction happens under -moz-column-content
anonymous block.
This patch excludes ColumnSetWrapperFrame from the debug check to meet
the expectation.
Add a reftest to make sure ::first-letter and ::first-line still work
after introducing ColumnSetWrapperFrame, and no assertion is fired.
Differential Revision: https://phabricator.services.mozilla.com/D20757
--HG--
extra : moz-landing-system : lando
Emilio Cobos Álvarez
Botond Ballo
Ciure Andrei
Miko Mynttinen
Timothy Nikkel
Masayuki Nakano
Andrea Marchesini
Brian Birtles
Hiroyuki Ikezoe
Glenn Watson
arthur.iakab
Jan Varga
hrdktg
Daniel Holbert
Dorel Luca
Jan Henning
David Major
Csoregi Natalia
Gurzau Raul
Matt Woodrow
shindli
Neil Deakin
Bogdan Tara
Jeff Muizelaar
Dan Glastonbury
Kartikaya Gupta
Geoff Brown
Tim Nguyen
Sam Sneddon
Chris Peterson
Makoto Kato
violet
Razvan Maries
Daniel Varga
Boris Chiou
Joel Maher
Narcis Beleuzu
Mike Conley
Lee Salzman
Cosmin Sabou
Boris Zbarsky
Jonathan Kingston
Ehsan Akhgari
Andreea Pavel
Brian Grinstead
Noemi Erli
Alex Henrie
Nika Layzell
Dzmitry Malyshau
Coroiu Cristina
Samuel Thibault
Mats Palmgren
Ting-Yu Lin