On Mon Jun 24 22:25:05 2024 +0000, Fabian Maurer wrote:
I reran the tests manually by uploading comctl32_test to the testbot, and there were no issues. I also restarted the pipeline by making a simple change, and now all is green.. I think that test failure isn't related to Arabic, it looks more like a spurious failure. Either the test I made is inherently flaky or it just ran into the timeout, although 200ms seemed plenty to me... I can't reproduce it locally either, even when running the test hundreds of times. I'd like to fix the test though, can you advice me on how to approach that?
Did you try with a larger timeout or Windows 10 21H1 in Arabic (Egypt) locale?
I would also try making the dialog foreground to ensure the input is not sent to other windows. If all that fails, you can add a flaky_if(in_arabic_locale).