Change the behaviour of PR creation #24
Conversation
We now do: - if the user closed the PR, we won't do anything as we assume the PR was rejected - if the PR is opened, we update the branch by force-pushing - if the PR was not there at all, we open a new one Fixes: stacklok#14
| @@ -225,6 +232,7 @@ func (fa *FrizbeeAction) createPR(ctx context.Context) error { | |||
| Username: fa.RepoOwner, | |||
| Password: fa.Token, | |||
| }, | |||
| Force: true, | |||
There was a problem hiding this comment.
Why the force here? Shouldn't a regular push work if the branch is new?
There was a problem hiding this comment.
To update the branch if an existing PR was opened. I'm fine removing this and I'm actually glad someone else than me or @rdimitrov joined the reviews, because we were discussing the workflow today and weren't sure what is more accepted by users.
Without force-push I guess the workflow would work like this:
- a PR is opened by this action
- someone pushes a new action into a workflow in a separate PR
- the original PR is merged
- on next tick of the action (either on schedule or on push to master) a new frizbee PR is created
With this patch:
- a frizbee PR is created
- another PR adds a new action
- the frizbee action is triggered either on schedule or on push to main and force-pushes to the branch, updating the existing PR
WDYT?
There was a problem hiding this comment.
Ah got it, thanks for the context. I thought you were trying to avoid touching any existing PRs. I got confused because I was expecting to see a rebase, and did not see the commit being recreated elsewhere.
SGTM
| if existingPR != nil && existingPR.GetState() == "open" { | ||
| fmt.Printf("PR %d already exists. Won't create a new one\n", existingPR.GetNumber()) | ||
| return nil | ||
| } |
There was a problem hiding this comment.
Since there is a force push before this, this block should probably be moved around line 213 to bail early and ensure NOOP.
There was a problem hiding this comment.
But then we wouldn't update an existing PR would we?
There was a problem hiding this comment.
Yeah thank for the clarification above!
We now do:
was rejected
Fixes: #14