@rsimha It's interesting that checks didn't catch this 🤔 we want to point it out just in case.
We were able to submit parseQueryString_ even though it was undefined.

Closing in favor of #34420

@rsimha It's interesting that checks didn't catch this we want to point it out just in case.
We were able to submit parseQueryString_ even though it was undefined.

@processprocess it looks like they both passed CI and were merged, but since neither touched the same lines it didn't trigger a merge conflict or force an update.

I really should have caught this since I was a reviewer on the original PR and the author on the other, but I likewise mistakenly assumed GH/CI would catch any errors that could arise

it looks like they both passed CI and were merged, but since neither touched the same lines it didn't trigger a merge conflict or force an update.

The crux of the issue is what Ryan said here. There's no good way to prevent this other than making sure that two PRs that could affect each other are tested and merged one after the other, and that PRs with older CI runs are rebased before they are merged.

In the case of #34372 and #34399, all the rules were followed, CI was green, and the PRs were merged a short time after running tests. Unfortunately, there was no merge conflict to flag a potential problem. The next line of defense is main branch builds, which worked as designed.