llama.cpp
llama.cpp copied to clipboard
Editing a PR description shouldn't cause a CI run
This seems excessive but I'm not sure how to turn that off (edited in build.yml?), without inadvertently also removing other triggers (such as a push to the branch the PR wants to merge).
Also, other actions such as requesting a review should not trigger a CI run IMO.
I have noticed this too. Makes me feel a bit bad every time that I modify the description, but am not sure if this has any real cost for us anyway.
build.yml
pull_request:
- types: [opened, synchronize, edited, reopened, review_requested, ready_for_review]
+ types: [opened, synchronize, reopened, review_requested, ready_for_review]
it was there so you can force a new CI run without pushing anything by editing the PR when it hangs due to github lagging but as you can also trigger that by other ways:
- requesting a new review
- mark as draft and back to ready for review
- close and reopen
it's probably a good idea to remove the 'edited' especially since the CI run is rather big now
On the other hand, draft pull requests skip most of the CI checks, that's not so good.
On the other hand, draft pull requests skip most of the CI checks, that's not so good.
I thought that usually drafts are incomplete and checks will likely fail, so I disabled them. But don't have very strong feeling about this - we can reenable if people prefer it
I think you would fix the checks while in draft state, then set it to "ready for review" once they pass. But other people may have different habits.
I realize I can enable actions on my own branch, but ideally I would like to have them run only on manual trigger - that doesn't seem possible.
I would suggest the following:
- disable the
edited, review_requested, ready_for_reviewcondition for pull requests - enable checks for drafts, so people get a chance for fixing failures before requesting reviews (yes, some checks may fail, but hopefully you're not hammering the git server with a thousand pushes, but test first on your machine)
- prevent merge with failing checks
I believe at least the last one would have to be set by @ggerganov in the repository settings.
Done
