Details
- Reviewers
mehdi_amini ldionne jhenderson
Diff Detail
- Repository
- rG LLVM Github Monorepo
Event Timeline
llvm/docs/GitHub.rst | ||
---|---|---|
40 | Is this an opportunity to squash? Does GitHub preserve better the history if we force-push on rebase but don't squash? The issue with this way of adding "fix up" commits is that a conflict during rebase may lead to as many redundant manual conflict resolution as there are "fix up" commits. |
llvm/docs/GitHub.rst | ||
---|---|---|
40 | Force pushes being unavoidable when doing rebases is one of the reasons I have a dislike of reviewing on GitHub, due to the issues it has. Regarding the additional sentence itself, I'm fine with it. I don't know of a way that will improve comment history more than anything else (squashing probably is safe, and I have no objection to encouraging it as per @mehdi_amini's comment). That being said, I haven't done any rigorous testing. |
llvm/docs/GitHub.rst | ||
---|---|---|
40 | Technically one would want to squash first, and rebase after! (to avoid the multiple conflicts resolution) There is also the option of merging main back into the user branch instead of force-pushing: why not this? |
llvm/docs/GitHub.rst | ||
---|---|---|
40 |
Doesn't that require a merge commit? I'm pretty sure we don't want those in our main branch history, and I don't know how that then plays with the squash & merge button at the end. |
llvm/docs/GitHub.rst | ||
---|---|---|
40 | With squash&merge and our branch protection, we can guarantee that main will end up with a single fast-forward commit, so it shouldn't matter there. |
Gentle ping on this. Let's either abandon this or merge it to clear up the review queue.
Is this an opportunity to squash? Does GitHub preserve better the history if we force-push on rebase but don't squash?
The issue with this way of adding "fix up" commits is that a conflict during rebase may lead to as many redundant manual conflict resolution as there are "fix up" commits.