Closed lonnen closed 1 year ago
It's possible to mark support for some versions as "experimental" so that the test outcomes don't contribute to the overall status of the build. It's not immediately obvious to me if this is useful for us. We might consider 3.11 experimental, but 3.10+ is failing and we really want to get to 3.10
https://docs.github.com/en/actions/using-jobs/using-a-matrix-for-your-jobs describes a configuring using "experimental" flag for specific targets in the build matrix. This is a generic capability that is being used with specific labels, like the word "experimental", but it doesn't show up in the UI very well. It's useful if we want to, say, have certain test fails cancel the whole run while other versions should report failures without interrupting the whole run but that's not particularly useful. I'd rather see all of the targets to establish a pattern than have this strange complexity beneath the UI.
There is some utility in having certain platforms marked as "experimental" but it is so buried in the config files that it seems hard to notice. I'm debating pulling it out to reduce the complexity, or adding a comment to the effect that the label is cosmetic and for documentation only.
merging this in without 3.10+ versions in order to expand coverage as much as we can immediately
Tox has only been wired up to run 3.9 in CI
This expands CI coverage to supported version of Python