I've been spending time on our Python-based test suites, to make it easier and
more "standardized" as we switch from `trunk` to `2.4.x`. The initial goal was
to combine the 2 different suites into one single one, with the same
architecture, but that was, frankly, way, way too difficult and had only
marginal advantages.
So I pivoted a bit with 2 new goals:
Centralize the 2 test suites via a common entry point
Have the same test suite(s) work as-is, byte-for-byte, on both `trunk` and
`2.4.x`. This meant enhancing the tests to be more aware of what system was
being tested and what modules/feature were backported, and what errors were
meaningful or could be ignored, due to the diffs in the codebase.
Both of these also affected the GitHub CI tests as well, which caused some
breakage as I was testing (see my reply about this).
Anyway, I think I have it now and the tests are now consistent and
self-contained.
--
Jim
"This is an outrage!"
Tony Harrison
