PR has caused desynchronization of master branch

Hi @tom I’ve just tried simulating the previous scenario and I’m not sure that the behaviour is entirely correct.

Please look at the following PR you can also find the specific build here.

It looks like it’s picked up the merge commit here, however the title for it seems to be based on actual commit that was merged itself [MISC] (deps): Bump haproxy in /internal/suites/example/compose/haproxy (#1359) as opposed to the message within the PR Merge branch 'master' into dependabot/npm_and_yarn/web/types/chai-4.2.13.

It also appears that the commit wasn’t picked up in the correct branch or Pulls section in codecov.

It is worthwhile mentioning that it hasn’t manifested in the same way, as in it doesn’t seem to count itself as a commit to master which is definitely an improvement, however it still does not appear completely fixed.

@nightah, thanks for this. I think we are getting closer to a resolution, but this helps dramatically. I’ll update here when we are able to push a fix.

@nightah, we made some changes to our bash uploader, would you mind seeing if this has been resolved?

@tom just tested it on the following PR, build can be found here.

The summary appears to be reporting the correct commit SHA’s parent to commit.

However I still am not seeing any of the commits showing up in the correct branch in codecov despite it knowing there was an update 9 minutes ago. It’s worthwhile calling out that it does appear that the pull in codecov has picked up the respective commits, so I’m not entirely sure if this is actually intended.

@nightah, thanks for this.

However I still am not seeing any of the commits showing up in the correct branch in codecov

Would you be able to run the bash uploader with the -v argument on? I’d like to take a look to see why the branches are incorrect

It’s worthwhile calling out that it does appear that the pull in codecov has picked up the respective commits

I don’t think I properly understood this. Are there older commits and/or newer commits that should have appeared? If older, that is probably intentional, but going forward, we should show the proper commits.

@tom I can turn it on shortly and provide another example.

For the second part of my last message I’m saying that despite the commits not showing up in the branches, they are showing up in the pulls/PR section per the link. The PR for that respective commit is all correct, including the merge commits and otherwise.

The part I’m unclear on is whether the commits are only meant to show up in one and not the other, for example only in pulls and not in branches?

@tom upon further inspection I don’t really see any of the branches other than master populated with results in codecov.

The only non-master branch I can find with results populated is the following recent one.

I’ve now turned on verbose mode for the bash uploader and you can find an updated PR (with merges per the earlier issues) at the following build location and here’s the github PR.

Here’s another example of the commits not showing up in branches from the actual PR to turn on verbose logging; PR and the associated build.

@nightah, this is indeed strange. It should exist on the other branch (not master). I’ll open up an issue with the product team, but thanks for your patience here.

@tom is it okay if I turn off debug logging? Are those examples that I have provided sufficient?

Hi @nightah, yes thank you so much!

Hi @nightah, we have made some improvements from our side here. Can you see if you still experience the issue?

Hi @tom, I’ve just ran into this, a PR was submitted from the main branch of a fork, codecov now seems to think that the commits on the forked branch are on the main repository which is now reporting low coverage (due to to failing tests)

PR: refactor(client): rename `Client` to `Prisma` by kivo360 · Pull Request #267 · RobertCraigie/prisma-client-py · GitHub
Codecov: Codecov
Badge

@RobertCraigie thanks for bringing this back up and for the links. Would you mind opening a new ticket for it?