ci/ci_commit_formatting_run: Problem with checking older MR branch. #12658
Conversation
dd8232f to
8f53143
Compare
|
Yep, this MR has failed the CI check in the same way: https://github.com/micropython/micropython/actions/runs/6475115234/job/17581470079?pr=12658 Note my current commit 8f53143 is not covered in that check at all, even though it should have failed. |
|
Code size report: |
Codecov Report
@@ Coverage Diff @@
## master #12658 +/- ##
=======================================
Coverage 98.39% 98.39%
=======================================
Files 158 158
Lines 20940 20940
=======================================
Hits 20603 20603
Misses 337 337 |
|
New commit 94fd52d updates the ci script to automatically fetch all upstream if the common commit hasn't been found; this should preserve the speed of CI checks for most MR's, but use slow & reliable if needed for older ones. It now passes that CI check (though it's the older version that only warned of missing sign-off) |
|
This MR has now been rebased to head to resolve merge conflicts. |
This commit updates the ci script to automatically fetch all upstream if the common commit hasn't been found; this should preserve the speed of CI checks for most PR's, and use a reliable but slow fetch if needed for older ones. Signed-off-by: Andrew Leech <[email protected]>
|
Thanks! |
I've noticed on a couple of MR's lately that the CI commit message check is failing, even when the MR commit looks fine.
One example of this is https://github.com/micropython/micropython/actions/runs/6059530353/job/16442749613?pr=12126
This an MR with 4 commits, however the CI check is looking at a long list of different commits:
I made this dummy MR branched from the same point as #12126 to test / examine the issue, though in testing locally I'm pretty sure I've found the problem. Assuming this initial push fails the same way, I'll push a follow up commit to check the solution.
This MR, like many other active MR's, is branched from an old enough point in master that when
ci_commit_formatting_runrunsgit fetch --depth=100 upstream masterit doesn't actually grab enough commits from upstream to find the common ancestor between the MR branch and master.It ends up checking only master... which includes some older commits that did not pass the check.