Add issue approver output and enhance approval logic - #206
Conversation
- Updated action outputs to include the GitHub username of the approver. - Modified approval logic to return the approver's username upon approval or denial. - Enhanced tests to validate the new approver output. Rename 'issue-approver' to 'issue-responder' for clarity in approval process
e4012c2 to
bac46d0
Compare
|
@lizziemac If possible, please review these changes; otherwise, I may publish the fork for my own use. thanks. |
Hi @felpasl Thanks a lot for this contribution. I understand that it has been very long, and you might have deployed your fork. The code changes look good to me. For good measure, is it possible to share some test results to show no breaking change for users operating with the previous version of inputs, and one test with your change? |
|
@felpasl Apologies for how long the back and forth is here, I wasn't getting emails for direct @'s. I can approve this once the test_docker target is separated out |
Making this change to merge it for now, later please create a new PR for these/similar changes.
|
Would need to refactor this quite a bit, will pick it up tommorow -- quite a lot of changes so not as straightforward as initial thought. |
CAOShurong
left a comment
There was a problem hiding this comment.
The exact head dbfab57 does not compile, so the new output cannot yet be exercised.
newCommentLoopChannelnow returnschan commentLoopResult, but the issue-state error path and externally closed issue path still send the integer literal1(main.go:147andmain.go:176).- The select arm binds
result, butmain.go:386,main.go:388, andmain.go:389still reference the removedexitCodevariable instead ofresult.exitCode.
I reproduced these five compile errors with go test -mod=mod -count=1 ./..., go vet -mod=mod ./..., go build -mod=mod ./..., and golangci-lint. The same commands pass on the current base. Public CI run 28285105918 likewise fails in Build and skips Test/Lint.
Please migrate both sends to commentLoopResult and use result.exitCode consistently, then rerun the focused/full tests and container action path. The approval, denial, closed-issue, and issue-responder output paths remain unverified until the exact head builds.
Review prepared with OpenAI Codex assistance; the exact head, commands, outputs, and public CI state were independently rerun and checked before submission.
related to #190