Closed dechamps closed 4 months ago
no, there are some branches like renovatebot/*
should not trigger ci.
And I think this is a special case that this repo need PR to non-master branch.
no, there are some branches like
renovatebot/*
should not trigger ci.
Why?
And I think this is a special case that this repo need PR to non-master branch.
That's not what my concern is though - it's not specific to that case. My concern is, if I push a branch to my fork in preparation to opening a PR, I don't get any CI results. I try to be a good citizen and only send you a PR once my branch passes CI, but your config is making it difficult to be a good citizen and then I have to waste your time opening PRs that don't even pass CI in the first place :(
Right now I'm working around that by patching in the config change in this PR in a separate branch, then build my change on top of that, then push that just so that I can run CI without bothering you with an incomplete PR. (Actually I realized just now that I can also work around that by opening a PR against myself… I guess that's a bit easier but still, ugh)
I don't know how you deal with that in your workflow - isn't it annoying when you push some branch and can't get any CI results without opening a PR first? It certainly is for me.
I'm not sure why this is controversial - CI is cheap, to me more CI is always better than less CI.
Why?
Beacuase renovatebot will create a branch in this repo then submit a PR, so CI will run twice.
That's not what my concern is though - it's not specific to that case. My concern is, if I push a branch to my fork in preparation to opening a PR, I don't get any CI results. I try to be a good citizen and only send you a PR once my branch passes CI, but your config is making it difficult to be a good citizen and then I have to waste your time opening PRs that don't even pass CI in the first place :(
Right now I'm working around that by patching in the config change in this PR in a separate branch, then build my change on top of that, then push that just so that I can run CI without bothering you with an incomplete PR. (Actually I realized just now that I can also work around that by opening a PR against myself… I guess that's a bit easier but still, ugh)
I don't know how you deal with that in your workflow - isn't it annoying when you push some branch and can't get any CI results without opening a PR first? It certainly is for me.
I'm not sure why this is controversial - CI is cheap, to me more CI is always better than less CI.
I really don't mind someone submit a PR and ci broken at first place. Moreover, you can submit a PR with draft
If this realy annoys you, you can merge this PR to your master branch, which will trigger ci of all branch if your repo.
Only running CI on master makes it harder to check the results without opening a PR.