Fetch, configure, and run node for Android on the test host, just like Linux tests do.
Make the node/HTTP/2 environment variables available to the tests on the device, and
use adb port forwarding to connect sockets. Finally, enable tests skipped for node.
Differential Revision: https://phabricator.services.mozilla.com/D60204
--HG--
extra : moz-landing-system : lando
It turns into a DOMException named 'TypeError' in the bindings, not into an actual TypeError.
Differential Revision: https://phabricator.services.mozilla.com/D60215
--HG--
extra : moz-landing-system : lando
"Scripts" wasn't making any sense for this marker and it was nearly imposssible
to understand. It should be "requestAnimationFrame callbacks" instead.
Differential Revision: https://phabricator.services.mozilla.com/D60229
--HG--
extra : moz-landing-system : lando
Done with:
./mach static-analysis check --checks="-*, modernize-concat-nested-namespaces" --fix .
and then clang-format on the files
Differential Revision: https://phabricator.services.mozilla.com/D58217
--HG--
extra : moz-landing-system : lando
Done with:
./mach static-analysis check --checks="-*, modernize-concat-nested-namespaces" --fix .
and then clang-format on the files
Differential Revision: https://phabricator.services.mozilla.com/D58217
--HG--
extra : moz-landing-system : lando
Optionally serialize N frames into a circular memory buffer, and save
them as part of wr-capture into tilecache/.
The new tile_view utility reads the folder and converts the frames to
svg for easier visualization, with a few extra features:
- color code each primitive based on which slice it is on;
- highlight new or moving primitives in red (brute force diff);
- print all invalidated tiles at the top and the invalidation reason;
- overlay the tile quadtree to visualize splitting & merging;
- HTML and JS wrappers for animation playback, timeline scrubbing;
Positioning of the tiles on the screen is a bit broken still; especially
during scrolling and "special" pages (like about:config).
Interning info is not used yet.
Differential Revision: https://phabricator.services.mozilla.com/D59975
--HG--
extra : moz-landing-system : lando
This is unaffected by the differences in font metrics between platform back-ends,
so should be safe to ship without significant risk that it'll be unstable or need
to be revised in future.
Depends on D59679
Differential Revision: https://phabricator.services.mozilla.com/D59680
--HG--
extra : moz-landing-system : lando
This is the main work of Richard Matheson's original patch, updated to current trunk code
and with the new attributes put behind prefs. Because some of the attributes may be more
stable than others (there was a move by Google to change how baselines are represented,
but then this was retracted because Safari is already shipping per the existing spec; and
we have some differences in how we handle font metrics between platforms which may affect
the font ascent/descent values), I've split this into several prefs so that we have the
possibility of enabling just the more stable (and/or more urgently requested) attributes.
(Note that this echos Google's approach per comment 30 of initially shipping part of the API.)
Depends on D59678
Differential Revision: https://phabricator.services.mozilla.com/D59679
--HG--
extra : moz-landing-system : lando
This is extracted from the original patch by Richard Matheson; it's not strictly necessary
to implementing the additional TextMetrics attributes, but OTOH if we're going to potentially
expose baseline attributes, it seems sensible to make them somewhat more meaningful than the
fallbacks in the current code.
Eventually we should use baseline tables from the font, but (a) we don't have support for
that in the back-end font code, and (b) very few fonts provide the data anyway, so while this
isn't perfect, for now it's the best we can do.
Differential Revision: https://phabricator.services.mozilla.com/D59678
--HG--
extra : moz-landing-system : lando
Currently, offset in root view in the test is copied from constants which are
computed from CSS rules. However, the point may be edge of each view and
that might be outside if it meets round error of layout. For making this
test more stable in any environments, this patch makes the test compute
point to fire wheel events dynamically.
Differential Revision: https://phabricator.services.mozilla.com/D60237
--HG--
extra : moz-landing-system : lando
Consider the following scenario:
1. Select a word in an iframe.
2. Select a word in the parent document. (The carets in iframe hide
due to blur, but the selection is still there.)
3. Scroll the parent document.
The selection in the iframe (made in step 1) has a non-collapsed range.
Thus the carets show again after scrolling due to `UpdateCarets()`
called in the end of `AccessibleCaretManager::OnScrollEnd()`.
To fix the dual AccessibleCaret in the same page (described in bug
1199967), we can simply show no carets if the window is not focused.
This behavior matches Google Chrome on Android.
Differential Revision: https://phabricator.services.mozilla.com/D52768
--HG--
extra : moz-landing-system : lando
Long-pressing on a text in an unfocused iframe to select a word never
works. Currently, you need to single tap to focus the iframe first.
Each PresShell has an associated AccessibleCaretEventHub. This patch
fixes this bug by routing the event to the AccessibleCaretEventHub under
the event point, and handle it there. If the event is not handled, then
we handle it by the focused AccessibleCaretEventHub as before.
I've experimented with only routing the event to the
AccessibleCaretEventHub under the event point, without routing to the
fallback focused AccessibleCaretEventHub. However, caret dragging didn't
work in iframes. I didn't debug further.
Differential Revision: https://phabricator.services.mozilla.com/D52767
--HG--
extra : moz-landing-system : lando
We want them to always appear on one line to avoid covering the toolbarbuttono,
even when there are multiple words in the badge.
Differential Revision: https://phabricator.services.mozilla.com/D60287
--HG--
extra : moz-landing-system : lando
Changes:
Reorganize, reorder and review the sets defined in test-sets.yml to be more conducive to future changes.
Also, migrate `mochitest-valgrind` to run with linux1804.
Reorganize:
Group similar sets together, separated by whether it is a feature test set or operating system test set.
Review:
Remove test sets that are no longer used.
Rename test sets to be more descriptive of what it contains.
Differential Revision: https://phabricator.services.mozilla.com/D60198
--HG--
extra : moz-landing-system : lando
Attempting to narrow down the crash of a null ServiceWorkerRegistrationInfo.
Differential Revision: https://phabricator.services.mozilla.com/D60326
--HG--
extra : moz-landing-system : lando
This also updates the addRDMTask testing function to take two new optional
promises for actions to take before and after invoking RDM mode. This test
uses that feature to set and check the resolution before and after RDM.
Differential Revision: https://phabricator.services.mozilla.com/D59822
--HG--
extra : moz-landing-system : lando
We need the resolution to be restored when and if the presshell creates
a MobileViewportManager. That decision happens in
PresShell::UpdateViewportOverridden. In order for the resolution to be
saved somewhere it has to exist outside the presshell itself, since
the presshell will be destroyed and recreated during some of the
operations that also destroy the MVM -- such as closing the devtools'
Responsive Design Mode. So Document holds this resolution and gets
accessor methods to allow the presshell to get and set that resolution
as needed.
Differential Revision: https://phabricator.services.mozilla.com/D59821
--HG--
extra : moz-landing-system : lando
Replace `isRestricting` with `getPriority` -- i.e., replace the binary restricting system with a general priority system. Instead of choosing restricting providers, the providers manager chooses the highest-priority providers.
Differential Revision: https://phabricator.services.mozilla.com/D60093
--HG--
extra : moz-landing-system : lando
This was happening because we had two exclude paths that were run through the 'collapse' algorithm:
editor/libeditor/tests/browserscope/lib/richtext
editor/libeditor/tests/browserscope/lib/richtext2
The problem was that in order to determine the base directory, the algorithm called
'os.path.commonprefix'. This function just returns the common string prefix, which
is the '.../lib/richtext' path. Even though in the above, the base *should* have
been '.../lib'.
To fix the problem, we add a check to ensure the computed base doesn't have any
sibling directories with the same prefix. If there are, it means 'commonprefix' was
too greedy.
Differential Revision: https://phabricator.services.mozilla.com/D60173
--HG--
extra : moz-landing-system : lando
Changes:
`mochitest-e10s` is ready to migrate to ubuntu1804 for all platform variants.
Failures are intermittent on non-fission runs.
Note that on `mozilla-central`, `linux64-ccov/opt` does not run `mochitest-e10s` and this behavior is carried over. Fission variant is also not run on `mozilla-central`.
Try push with fission and ccov is included for sake of completeness.
Differential Revision: https://phabricator.services.mozilla.com/D59970
--HG--
extra : moz-landing-system : lando
