Hint at lack of access as a possible reason to error

[MichaelChirico]

As discussed internally, there might be some rare cases where jj tries to update to an error lacking permissions for the current user at the current time.

[MichaelChirico]

Unsure what to do with this:

TypeError: Cannot read properties of undefined (reading 'state')
Error: Unhandled error: TypeError: Cannot read properties of undefined (reading 'state')
    at eval (eval at callAsyncFunction (/home/runner/work/_actions/actions/github-script/v7.0.1/dist/index.js:35424:16), <anonymous>:33:20)
    at process.processTicksAndRejections (node:internal/process/task_queues:95:5)
    at async main (/home/runner/work/_actions/actions/github-script/v7.0.1/dist/index.js:35522:20)

[martinvonz]

Unsure what to do with this:

I also have no idea what that is about. Maybe it's something temporary. Try rerunning the check if you haven't already.

[MichaelChirico]

Hm, possibly it's an action that can't work on commit to forks? I've encountered that elsewhere...

[martinvonz]

Hm, possibly it's an action that can't work on commit to forks? I've encountered that elsewhere...

Oh, good point. @icorbrey: I think this is a happening in the check you added. Do you have time to look into it?

[MichaelChirico]

No JS expert (or even novice, really) but with Gemini assist + poring over https://github.com/jj-vcs/jj/blob/main/.github/workflows/pr.yml I can reason that:

• I have not correctly formatted my commit descriptions (oops)
• However there are no githubActionsBot reviews, therefore latestReview is empty / doesn't have a .state attribute

Thus we reach here and see the error:

jj/.github/workflows/pr.yml

Line 51 in 1e58ca2

if (latestReview.state !== 'CHANGES_REQUESTED') {

I think I can reverse engineer how to fix it in this PR by getting out of the badCommits.length>0 branch.

I can have a go at fixing the GHA in a separate PR.

[MichaelChirico]

OK, that item is ✅ now

[icorbrey]

Huh, didn't run into that with my testing. Sorry about that!

[martinvonz]

Looks like we forgot to review this. It looks good to me but I'm not allowed to approve. Can someone else (review and) approve it?

[MichaelChirico]

help -- I'm in git hell. I don't know how to handle the merge commit that's here now.

[PhilipMetzger]

I know this video has been helpful to others struggling with Git: https://youtu.be/ElRzTuYln0M otherwise consider using jj on Git instead of Piper for this :-).

[MichaelChirico]

Thanks for the video -- TIL about fixup, that's nice.

Unfortunately it didn't really help in my case, I think the issue was ultimately that I'd forgotten to pull the merge commit I made in the browser from my branch (argh) so it wasn't showing up locally.

I then managed to bork things even more thoroughly a few times -- the joys of git :)

I think it's finally fixed 🤞

[MichaelChirico]

Hi all, friendly ping :)

[martinvonz]

The PR is already approved so you should be able to merge it. Or is it still blocked by failing CI? You can try re-running the Windows tests.

[MichaelChirico]

Oh, thanks. I saw the Approval but it was before two more reviewer comments, I wasn't sure whether to wait for follow-up from those reviewers before shipping.