I experienced unexpected behavior when working with this pull request.
I had pushed this commit to the relevant branch, which I recall having triggered your CI process; however, I then forcibly (non-fast-forward) pushed another commit to that branch almost immediately thereafter (an ancestor), which apparently canceled the first CI job and started a second. But, I didn’t stop there! I pushed that original commit (84ad97fe…) back to the branch, yet it neither canceled that second job in order to re-initiate the first, nor canceled that second job in order to create a third. Instead, it pulled up the cached results for the first job: “X”; failed; canceled.
Well, I did the silly thing: I closed the “errant” pull request, and opened another; naturally, the cached result still appeared. So, I altered the commit hash (by modifying the commit object) and then forcibly pushed it, thereby resolving the issue, as this triggered a successful CI process.
However, I later noticed that the closed pull request also ended up showing success with the original commit, so I could have just waited for it, I guess; if that be the case, then I suggest the following:
- Notify people (including on GitHub) that a new job will be started soon, or provide a button to initiate a re-try manually.