On 2023/01/08 11:52:35 +0100, Stefan Sperling <stsp@stsp.name> wrote:
> I found a situation where 'got rebase' will proceed even though the
> work tree is out-of-date. This happens because a seqence of youngest
> common-ancestor and same-branch checks will trivially succeed, leading
> 'got rebase' to assume that this rebase is a fast-forward. However,
> it only looks like a fast-forward because we miss the fact that the
> work-tree's base commit is out of date, ignoring newer commits on the
> work-tree's branch.
> 
>  This situation is not covered by the existing rebase_out_of_date test.
> To reproduce the problem, run the new test added with this patch, with
> the got/got.c part of this patch backed out:
> 
> test_rebase_out_of_date2 rebase succeeded unexpectedly
> test failed; leaving test data in /tmp/got-test-rebase_out_of_date2-N3qmgXLx
> $ cat /tmp/got-test-rebase_out_of_date2-N3qmgXLx/stdout
> refs/heads/newbranch is already based on refs/heads/master
> Switching work tree from refs/heads/master to refs/heads/newbranch
> U  epsilon/zeta
> U  gamma/delta
> Updated to refs/heads/newbranch: f5fd50fc48b880a976504bb432c60f1372c5dc9c
> $
> 
> ok?

ok op@