Branch: refs/heads/main
Home: https://github.com/WebKit/WebKit
Commit: f2815ca5349748843abc9a3c2b3b675b359df489
https://github.com/WebKit/WebKit/commit/f2815ca5349748843abc9a3c2b3b675b359df489
Author: Wenson Hsieh <[email protected]>
Date: 2024-01-17 (Wed, 17 Jan 2024)
Changed paths:
M LayoutTests/editing/input/ios/typing-with-inline-predictions-expected.txt
M LayoutTests/editing/input/ios/typing-with-inline-predictions.html
M
LayoutTests/editing/selection/ios/tap-to-change-selection-after-accepting-inline-prediction.html
Log Message:
-----------
[iOS] A couple of inline prediction layout tests are flaky
https://bugs.webkit.org/show_bug.cgi?id=267639
rdar://121118748
Reviewed by Megan Gardner.
Attempt to triage a couple of flaky layout tests that exercise inline text
predictions. I wasn't
able to reproduce any test failures when running either of these (running them
fully parallel for
several hundred iterations, or when running them in order along with the rest
of the test group for
several dozen iterations).
These are purely speculative adjustments to the flaky tests.
* LayoutTests/editing/input/ios/typing-with-inline-predictions-expected.txt:
* LayoutTests/editing/input/ios/typing-with-inline-predictions.html:
Instead of only waiting for `typeCharacter`, wait for the editing container to
receive an `keyup`
event as well before trying to send the next set of key events.
Also, temporarily add a 100 ms delay between keystrokes. If this patch fixes
the flakiness observed
in our automation, we should try removing this delay and see if these tests
start flaking again. If
so, we likely have another source of flakiness when synthesizing key events,
which the above change
to wait for the `keyup` event does not address.
Lastly, add a `console.log` here so that if the test begins to time out due to
never observing an
`keyup` event, we can easily tell.
*
LayoutTests/editing/selection/ios/tap-to-change-selection-after-accepting-inline-prediction.html:
Same as above, but additionally remove the `allowsInlinePredictions=true` test
option. This test
doesn't actually need this option, because it explicitly sets inline
predictions; I think it's
possible that, by enabling inline predictions, the real system predictions (or
lack thereof)
override the custom prediction ("I want to c<elebrate>").
Canonical link: https://commits.webkit.org/273149@main
_______________________________________________
webkit-changes mailing list
[email protected]
https://lists.webkit.org/mailman/listinfo/webkit-changes
