Test suite should post-process failures
While we would like our test suite to "Just Work", this is never the case. Not only some test cases break, but we also have some kind of failure that can happen mostly everywhere, and that we learn to ignore as hiccups:
- Frequent htpdate failure in local test suite run (#18529)
- Test suite breakage when restoring an old Bulls... (#18953 - closed)
We'd like to have a mechanism which, in case of failures, applies rule to better identify this failure and automatically provide useful data to the user. For example, "Guest disabled display", or cannot find GnomeApplicationsMenu.png
could be considered a "skipped" test. It needs to be rerun, but is not a real failure.