Branch: refs/heads/main
Home: https://github.com/WebKit/WebKit
Commit: 5273909547b13c75ce88e806b773737baa363c9d
https://github.com/WebKit/WebKit/commit/5273909547b13c75ce88e806b773737baa363c9d
Author: Sam Sneddon <gsnedd...@apple.com>
Date: 2023-10-18 (Wed, 18 Oct 2023)
Changed paths:
M LayoutTests/TestExpectations
M
LayoutTests/http/tests/security/contentSecurityPolicy/upgrade-insecure-requests/basic-upgrade-cors.https-expected.txt
M
LayoutTests/http/tests/security/contentSecurityPolicy/upgrade-insecure-requests/basic-upgrade-cors.https.html
M
LayoutTests/imported/w3c/web-platform-tests/cookies/partitioned-cookies/partitioned-cookies.tentative.https-expected.txt
M
LayoutTests/imported/w3c/web-platform-tests/html/browsers/browsing-the-web/navigating-across-documents/source/navigate-child-function-parent-then-fragment-expected.txt
M LayoutTests/platform/wk2/TestExpectations
M LayoutTests/resources/testharnessreport.js
Log Message:
-----------
[testharnessreport.js] Only pay attention to the top-level completion handler
https://bugs.webkit.org/show_bug.cgi?id=259409
rdar://problem/112683033
Reviewed by Jonathan Bedard.
Previously, we were considering the test as having to run to completion
when any testharness.js completion handler first ran (which was
sometimes that of the frame within the opened window; the nondeterminism
here just adds flakiness to the already bad behaviour).
Instead, only output anything for the top-level completion handler, as
all results should be passed up to it. Note wptrunner with the WebDriver
or Marionette executors only ever pays attention to the top-level
completion handler (as they only pay attention to the top-level frame &
window), thus they don't have any flakiness like this.
Ideally testharness.js would have an API we could use for this; I've
filed an RFC at https://github.com/web-platform-tests/rfcs/pull/168 for
this, but there's no reason not to do the simple fix ourselves, as at
least for now this is a strict progression. (It would stop being a
strict progression if at some point testharness.js started allowing
fetch_tests_from_window() with a window with noopener.)
It seems likely we have other tests that are marked as flaky because of
it, but alas there's no easy way to find what tests will have been fixed
by this.
* LayoutTests/TestExpectations:
*
LayoutTests/http/tests/security/contentSecurityPolicy/upgrade-insecure-requests/basic-upgrade-cors.https-expected.txt:
*
LayoutTests/http/tests/security/contentSecurityPolicy/upgrade-insecure-requests/basic-upgrade-cors.https.html:
*
LayoutTests/imported/w3c/web-platform-tests/cookies/partitioned-cookies/partitioned-cookies.tentative.https-expected.txt:
*
LayoutTests/imported/w3c/web-platform-tests/html/browsers/browsing-the-web/navigating-across-documents/source/navigate-child-function-parent-then-fragment-expected.txt:
* LayoutTests/platform/wk2/TestExpectations:
* LayoutTests/resources/testharnessreport.js:
Canonical link: https://commits.webkit.org/269483@main
_______________________________________________
webkit-changes mailing list
webkit-changes@lists.webkit.org
https://lists.webkit.org/mailman/listinfo/webkit-changes
