tree 4b38211bb20105e8e7510d87b79eb8e9fea3f632
parent 71200b77aa8dbd4b13388ba6e5a797d0dd72f1d6
author Domenic Denicola <domenic@chromium.org> 1709530551 -0800
committer Blink WPT Bot <blink-w3c-test-autoroller@chromium.org> 1709531723 -0800

Re-do close watcher user activation tests

The tests were not accurate, because multiple tests in the same file
would result in leftover user activation, corrupting the results.

Additionally, the tests were not exhaustive enough to catch the cases
discussed in https://crbug.com/41484805. Furthermore, the initial
attempt to fix those cases, in
https://chromium-review.googlesource.com/c/chromium/src/+/5187183,
broke our desired invariant, that the number of Android back button
presses/gestures to escape the current page would be at most the
number of user interactions + 2: it made it possible to trap the user
and require infinite back presses/gestures.

The new tests are sufficiently exhaustive and systematic. The test
expectations added here show that we are not currently sending some of
the cancel events that we could, i.e. that issue 41484805 is not
fixed. A future CL will make all the tests pass.

Another bug discovered in this whole process is that the grouping
mechanism interacts poorly with destroying close watchers. Because "is
grouped with previous" is tracked on the close watcher, once the
close watcher is destroyed, groups can get collapsed together. This is
illustrated by the new nynn-destroy.html test failure. A future CL
will fix it.

Bug: 41484805
Change-Id: Iec4ee5f195640cbae60a3b1aa6127cb0f9cf7ba0
Reviewed-on: https://chromium-review.googlesource.com/c/chromium/src/+/5318859
Reviewed-by: Joey Arhar <jarhar@chromium.org>
Commit-Queue: Domenic Denicola <domenic@chromium.org>
Cr-Commit-Position: refs/heads/main@{#1267687}