Jonas Greitemann

After updating from GCC 13.2 to 14.1, the false-positive warning is indeed gone. This issue can probably be closed as this confirms that this is a compiler regression?

Hi @shikanime, given that Martin has approved, would you give it one last rebase to resolve the conflicts? It should be fine to merge then, in time for the 0.36...

Easier in the sense that RepoPath uses `/` on either platform? Yes, unless we ever want to support completion on Powershell at some point. Idk, maybe slashes would be permissible...

@yuja, I went with your suggestion and switched to `RepoPath`. Took me a moment to realize, but this is actually much better on Windows as it allows mixing and matching...

As of #6411, this no longer produces a panic. However, the behavior is still faulty: Following the above repro and not selecting any changes in the diff editor produces a...

Exactly right. The existing `test_edit_diff_builtin*` tests already merely check that the left/right tree is recovered when selecting no/all changes in the diff. Fuzzing for this same property would just generate...

@thoughtpolice, since you've also reviewed #6316 on behalf of the maintainers, would you be willing to take a look at this one, too?

I've just "reheated" the PR. The new test correctly found that the schema default for `git.write-change-id-header` was missed in #6475. I've added a change to fix that, too.