ci: reintroduce macOS and Windows testing on the master branch
#1116
Conversation
JKRhb
changed the title
master branchmaster branch
Codecov ReportAll modified lines are covered by tests ✅
Additional details and impacted files@@ Coverage Diff @@
## master #1116 +/- ##
=========================================
Coverage ? 75.67%
=========================================
Files ? 80
Lines ? 16159
Branches ? 1520
=========================================
Hits ? 12229
Misses ? 3888
Partials ? 42 ☔ View full report in Codecov by Sentry. |
danielpeintner
left a comment
danielpeintner
left a comment
There was a problem hiding this comment.
LGTM
The only use-case I have in mind is the following. At the moment MacOS and Windows are run only against master. Good so far, However, how can we (even manually) trigger the CI to run against all OS'es ... because we want to check that also in some special cases...
|
I think with the 
|
|
Hmm. On second thought, that probably does not change the logic within the CI matrix. Maybe there could be another case added for that? |
Not entirely sure if it works, but if it does, the last commit should have added it in ;) |
|
I'm not entirely convinced about narrowing down the matrix to only work on merge. I would like to know immediately if a PR is breaking and fix the errors in the PR. I appreciate the effort, but my preference would be to wait longer and be sure that we have all greens. |
Yeah, I think we can definitely go this route, as mentioned in the issue, the total time is now about 8 minutes which should be fine as, if something should be completely broken, we will notice it much earlier on |
|
I just opened #1121 as alternative :) |
Yes, maybe that's better... |
3 participants
As discussed in #882, this PR reintroduces automated testing with Windows and macOS on the
masterbranch. To limit the use of macOS and Windows to themasterbranch, I adjusted the testing matrix so that the two OSs are excluded if the current branch should not bemaster. This rendered the other workflow definition file obsolete, which is why I removed it.While I was at it, I slightly adjusted the
onsection in the workflow file so that the workflow only runs in thepushcase when it is a push to themasterbranch. This resolves the slight annoyance that the CI ran twice for some reason in the case of a PR coming from a fork.With this PR, we could consider #882 resolved. However, we could also leave it open and evaluate if macOS and Windows testing should be reintroduced for PRs as well.