https://bugs.freedesktop.org/show_bug.cgi?id=99649
--- Comment #3 from Tomi Sarvela <[email protected]> ---
I was asked to explain my issue a bit better, so here's by pov:
Catching the exception is clear. The issue is known and easily hit, and
non-catched exception is a bug.
Choosing what to do:
Continue with Warning
- This helps CI: testlist might be maintained locally, and removing test
upstream breaks every run after update at the moment
- This helps user: warning about missing test, but test anyway what you can
- Test order is in meta/tests, just as with regexes
Abort with Error
- User: makes it clear that testlist can't be run as-is. Force to edit testlist
- CI: breaks. Resolution: don't build piglit from upstream automatically
- Test order can be read from testlist, in addition to meta/tests
One possibility is to choose the behaviour with a commandline option. I
wouldn't mind that, either.
I don't think testlist are very widely used yet, so choosing the correct
default is possible. I'm liking "Continue with Warning", because that's the
unix way.
--
You are receiving this mail because:
You are the QA Contact for the bug.
_______________________________________________
Piglit mailing list
[email protected]
https://lists.freedesktop.org/mailman/listinfo/piglit
