Windows programs run via Wine don't like Unix absolute paths (they look
like command line arguments), so we need to use relative paths.
Mingw already run fxc2 via wine, but for some reason it doesn't care
about the Unix absolute paths. genshaders does need some adjustements to
run properly with the real fxc.
Now, on actual Windows, because the temporary directory where
tempfile.NamedTemporaryFile creates files by default is not necessarily
on the same drive as where the command runs from, a relative path can't
be constructed. So we also force the temporary file to be created in the
current (obj) directory.
There is no similar concern for other files because we only go from
objdir to srcdir, and the build system already doesn't support both
being on a separate drive.
While here, flush stdout when the genshared script writes to it, so that
the messages are printed out immediately rather than randomly, later,
after output from subprocesses.
Differential Revision: https://phabricator.services.mozilla.com/D64294
--HG--
extra : moz-landing-system : lando
Inputting Unicode scalar value with `Alt` + numpad keys is available even
without NumLock state. However, `Alt` + function key on numpad should also
be worked because user may intent to perform it. Therefore, this patch
stops the hacking for bug 1606655 when given virtual keycode value is a
function key, but this means that users cannot type a Unicode scalar value
without NumLock key state if the value includes `7` (`Home`),
`4` (`ArrowLeft`), `6` (`ArrowRight`) because Firefox UI handles they are
shortcut keys (for "Go home", "Go back" and "Go forward"). Unfortunately,
I have no idea how to solve this conflict (if it's second key or latter key
after pressing `Alt` key, we could do that with a boolean flag, but I don't
like to make it only for this kind of edge case unless a lot of users want
to do it).
Differential Revision: https://phabricator.services.mozilla.com/D63782
--HG--
extra : moz-landing-system : lando
Automatic update from web-platform-tests
Add a test for content-dpr in various scenarios (#21962)
* Add a test for content-dpr in various elements
* Fix typo
--
wpt-commits: a29d8d7a84b904431727039d9db9992cc77288ac
wpt-pr: 21962
Automatic update from web-platform-tests
cc Adjust html-aria/presentation-role alt values
--
wpt-commits: 1fc6e4c3325792b5cbab0083a1e68562503bd677
wpt-pr: 21977
Automatic update from web-platform-tests
[css-grid] Fix referencing name of 1st line followed by auto repeat()
When referencing a line name in the grid-placement properties, if the
first track was defined using the auto repeat() syntax, then the code
would assume that the line had to appear after the repeated tracks.
However, it can precede them, e.g.
grid-template-columns: [first] repeat(auto-fill, 10px)
This patch fixes this by removing the special casing for auto repeat()
being the first tracks.
BUG=966090
TEST=external/wpt/css/css-grid/placement/grid-placement-using-named-grid-lines-002.html
Change-Id: Ic9c1199532f8ed06c958b80aef4580b7a318490c
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2071873
Reviewed-by: Javier Fernandez <jfernandez@igalia.com>
Commit-Queue: Oriol Brufau <obrufau@igalia.com>
Cr-Commit-Position: refs/heads/master@{#744426}
--
wpt-commits: a4677f682d251f7c072772973058e77777711564
wpt-pr: 21964
Automatic update from web-platform-tests
Improve python3 support of test_wpt (#21838)
Main change is that git commands return Text now instead of bytes. Updated callsites and also mypy comments.
--
wpt-commits: b4023970550ce0d454e700fad25e957ad1366ceb
wpt-pr: 21838
Automatic update from web-platform-tests
Rewrite self.html as an any.js test.
Bonus: this adds coverage for shared and service workers.
--
wpt-commits: 342de366ef12e8daded996654cc4b05f8883ff2d
wpt-pr: 18228
Automatic update from web-platform-tests
update broadcastchannel test to reflect closed check move into queued task.
--
wpt-commits: 691a6a903c5783914efe4bd8657967ac8f1ea0b9
wpt-pr: 21895
Automatic update from web-platform-tests
Clear-Site-Data must uncontrol affected service worker clients
Adds an immediate flag to unregister jobs for service worker
registrations. When set, the immediate flag enables unregister jobs to
run the "Clear Registration" algorithm immediately without waiting for
the active worker's controllees to unload. After an immediate clear,
the registration's controllees become uncontrolled. The controllees
receive the 'controllerchange' event when the
navigator.serviceWorker.controller is set to null.
This change updates ServiceWorkerContextCore::DeleteForOrigin() to set
the immediate unregister job flag to true. All other callers set the
immediate flag to false. The storage quota manager uses
ServiceWorkerContextCore::DeleteForOrigin() when clearing data for the
origin. In addition to evicting origins to free up storage quota, this
code path is used by scenarios like the Clear-Site-Data network
header and the "clear browsing data" privacy UI setting.
The change adds a new function to the ServiceWorkerRegistration class,
DeleteAndClearImmediately(), which is used by ServiceWorkerUnregisterJob
when the immediate flag is true. DeleteAndClearImmediately() is very
similar to the existing function, ClearWhenReady(), but it does not wait
to clear the registration when the active worker has controllees.
ServiceWorkerContextCore::DeleteForOrigin() is also updated to call
DeleteAndClearImmediately() on the origin's uninstalling registrations.
Registrations stuck in the uninstalling state are waiting for the active
worker's controllees to unload.
ServiceWorkerContextCore::DeleteForOrigin() must stop waiting and clear
the registration immediately. GetUninstallingRegistrationsForOrigin()
is added to the ServiceWorkerRegistry to enable
ServiceWorkerContextCore::DeleteForOrigin() to find the uninstalling
registrations it needs to clear.
For testing, the change adds a new unit test that sets the immediate flag
to true in ServiceWorkerJobTest. The change also adds a new service
worker WPT test, unregister-immediately.https.html. The new WPT test
uses the Clear-Site-Data network header to unregister service workers
with the immediate flag set to true. The WPT test adds a test case for
each service worker state. For example, one test case uses
Clear-Site-Data to clear a registration with a service worker in the
'parsing' state. Other test cases cover the other service worker states
like 'installing', 'installed', 'activating' and 'activated'.
Additional test cases verify controllee state, which includes verifying
what happens to pending fetch events when the controller clears.
Bug: 1014114
Change-Id: I3baee1117e6dbd349ec4c98c796d4e1b2bddde96
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2040533
Reviewed-by: Matt Falkenhagen <falken@chromium.org>
Reviewed-by: Makoto Shimazu <shimazu@chromium.org>
Commit-Queue: Steve Becker <stevebe@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#743477}
--
Change two service-workers/ tests to use promise_rejects_{dom,js}
--
wpt-commits: d5d27547fd654c8f70403cec0fa34f527060272c, fe4c8a394bf19a480354748fb11db2e35df6e637
wpt-pr: 21620
Automatic update from web-platform-tests
[v8][wasm] Fix WebAssembly streaming tests
This CL fixes a test that is checking the wrong behavior at the moment,
and disables the test until the fix in V8 lands.
Background:
The idea of the test is to create an invalid WebAssembly module, and
then check that the module is rejected by streaming compilation. The
invalid module was created by taking an empty module and adding two '0'
bytes at the end. However, two '0' bytes represent an empty unknown
section, which was originally accected by the streaming decoder. The
test, however, misinterpreted the test result (i.e. "FAIL Invalid code:
compileStreaming assert_unreached: Should have rejected: undefined
Reached unreachable decoder" a.k.a. FAIL, streaming compilation
succeeded unexpectedly) as the error message of the streaming compiler,
and then marked it in the test expectations as the expected behavior.
The CL changed the test now to add bytes that clearly create an invalid
module, and also deleted the invalid test expectations.
R=leszeks@chromium.org
Bug: v8:10126
Change-Id: Ie49a7a890abd841a85dc9f634308762ea07e5928
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2066907
Reviewed-by: Leszek Swirski <leszeks@chromium.org>
Commit-Queue: Andreas Haas <ahaas@chromium.org>
Cr-Commit-Position: refs/heads/master@{#743908}
--
wpt-commits: 95e12101852ce37054603845f49056cc9ec0da0a
wpt-pr: 21931
Automatic update from web-platform-tests
Align <pattern> and <*Gradient> attribute collection
Currently we stop collection of the pattern attributes if one of the
<pattern>s that is inherited from is not in the layout tree (but still
use the result). For gradients this cause the generation of the paint
server to fail (triggering fallback if available).
So consider a missing LayoutObject the same as a missing element in the
various Collect...Attributes implementations, essentially ignoring the
reference.
This also aligns our behavior with that of Gecko.
Add const qualification to the CollectGradientAttributes() methods.
Bug: 109212
Change-Id: Ie8a22664c4dc37ebc10fb02f18e1176a9d41a0e0
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/968924
Reviewed-by: Philip Rogers <pdr@chromium.org>
Commit-Queue: Fredrik Söderquist <fs@opera.com>
Cr-Commit-Position: refs/heads/master@{#743900}
--
wpt-commits: 45d0e5e00724e4a3bf10b676d09d5b93c814e622
wpt-pr: 21919
Automatic update from web-platform-tests
Split html/.../query-encoding/ tests (#21881)
These tests cover 5 encodings (UTF-8, UTF-16LE/BE, Windows-1251/1252)
and exercise many features (HTML, CSS, SVG, Workers, ...). They're
currently all skipped in Chrome (and other browser CI systems), since
something in that huge matrix fails or times out for everyone.
https://github.com/web-platform-tests/wpt/issues/4934 suggests
splitting these into separate files. As a stopgap, split them out
using the WPT "variant" mechanism and provide expected failure files
for subsets, which should make fixing the tests and/or browsers more
actionable, and let us catch new regressions.
Bug: 930297
Change-Id: I4813281f547216b2b5f377118a58a419f62cfbc7
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2063808
Commit-Queue: Philip Jägenstedt <foolip@chromium.org>
Reviewed-by: Philip Jägenstedt <foolip@chromium.org>
Cr-Commit-Position: refs/heads/master@{#743019}
Co-authored-by: Joshua Bell <inexorabletash@gmail.com>
--
wpt-commits: d40812425800cccad65582ad1132d97303f1f1b9
wpt-pr: 21881
Automatic update from web-platform-tests
Fetch HTTP(S) scheme/data url scripts from data url iframe/dedicated worker/shared worker (#21877)
This pull request adds web-platform-tests to check if fetching scripts from data url iframe, dedicated worker or shared worker is appropriately allowed or rejected by CORS.
--
wpt-commits: 2be2d7e3abcde5baded3448b85d0bb88e58d3cf7
wpt-pr: 21877
Automatic update from web-platform-tests
Implement dirname for form submit and directionality for element
--
Move dirname collection to input
--
Rearrange directionality algorithm functions
--
Simplify node#parent_directionality
--
wpt-commits: e5321c5c1123b419f9c871bee7c13354b60ab911, 4bd794f7c273788e8a8842ca3c5ee3fd0d9b13b3, cee8a3e8b1b97e6cc1ae61c4625544cbf737ff99, 0350340c01766a08efd65abba28ffeff64862842
wpt-pr: 21932
Automatic update from web-platform-tests
Fixed initialization of animation hold time when scroll timeline is inactive
https://chromium-review.googlesource.com/c/chromium/src/+/1945923
change erroneously initializes hold_time_ to null when scroll
timeline is inactive. This makes assert to fail in
Animation::CommitPendingPlay at:
DCHECK(start_time_ || hold_time_);
Bug: 1042924
Change-Id: I98060d065da93bb633d6cbc1506c591194a11692
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2037348
Commit-Queue: Olga Gerchikov <gerchiko@microsoft.com>
Reviewed-by: Majid Valipour <majidvp@chromium.org>
Cr-Commit-Position: refs/heads/master@{#744037}
--
wpt-commits: 9b38d0ac614a4132804b076c461738871689fae7
wpt-pr: 21585
Automatic update from web-platform-tests
[FlexNG] Don't let min size go negative
We were subtracting border_padding when it wasn't present if the item
had min-size:auto but didn't qualify for the special flex min-size:auto
behavior because, for instance, it has overflow: hidden.
This fixes the telemetry_perf_unittests failure FlexNG was hitting on
the bots, which was caused by a recorded google docs page.
Change-Id: Ia6bde490651daf63e4ca77f9525870987bb9fe5c
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2065626
Commit-Queue: David Grogan <dgrogan@chromium.org>
Reviewed-by: Ian Kilpatrick <ikilpatrick@chromium.org>
Cr-Commit-Position: refs/heads/master@{#744017}
--
wpt-commits: 2ee59347f0e4e93b29cecd8e87e8d1d509391774
wpt-pr: 21893
Automatic update from web-platform-tests
Make Resize Observer notification process stateless
Remove SetNeedsResizeObserverUpdate, and associated flags and functions
in order to make the notification delivery process stateless. With this
change, ResizeObservation::ObservationSizeOutOfSync will determine
whether notifications should be delivered rather than
element_size_changed_ and observer_changed flags being set by
SetNeedsResizeObserverUpdate, which is called at various points in the
layout process.
Cases involving many ResizeObservers were considered, and it is decided
that it would be more desirable to have slower pages which contain
many (>10^4) observers rather than slowing down
LayoutBox::StyleDidChange with extra operations (which would have been
required for scenarios involving content box changes).
With this change, content box changes can be observed, fixing all WPT
observe.html tests.
Bug: 1034329
Change-Id: Ie2225ad2865ee15d82be548c1455949946734074
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2033849
Reviewed-by: Philip Rogers <pdr@chromium.org>
Reviewed-by: Daniel Libby <dlibby@microsoft.com>
Commit-Queue: Sahir Vellani <sahir.vellani@microsoft.com>
Cr-Commit-Position: refs/heads/master@{#744016}
--
wpt-commits: a7ad7389eacd18b6ad2bbcddae7108955302d1e4
wpt-pr: 21765
Automatic update from web-platform-tests
[PerformanceObserver] Fix observe() edge cases
Currently, two calls to observe with the same 'type' will result in a
console error, whereas the spec implies that this is valid, with the
later call overriding the initial one. This CL fixes that. It also fixes
a problem where disconnect would not clear the filter_options_, as shown
in https://github.com/w3c/performance-timeline/pull/161. This would not
be a problem with 'entryTypes' because this variable would be replaced
on every call to observe(). Tests are added for these two cases.
Bug: 1052509
Change-Id: I75865b2925e2fb1a7e7fd64d7e979907c3b7650c
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/2067250
Reviewed-by: Yoav Weiss <yoavweiss@chromium.org>
Commit-Queue: Nicolás Peña Moreno <npm@chromium.org>
Cr-Commit-Position: refs/heads/master@{#744012}
--
wpt-commits: fcbabacaabd8a0e473ba874ec2207390c61bed94
wpt-pr: 21920
Ciure Andrei
Mike Hommey
Andrew McCreight
Andrea Marchesini
Bogdan Tara
Hiroyuki Ikezoe
Masayuki Nakano
Cosmin Sabou
moz-wptsync-bot
Noam Rosenthal
Michael[tm] Smith
Oriol Brufau
Sergio
Ms2ger
Josh Matthews
Gregory Terzian
Steve Becker
Andreas Haas
Fredrik Söderquist
Blink WPT Bot
Eriko Kurimoto
elkurin
autofoolip
Dmitry Kolupaev
Olga Gerchikov
David Grogan
Sahir Vellani
Nicolás Peña Moreno