On Wed, Feb 05, 2025 at 04:04:47PM -0800, Junio C Hamano wrote: > Junio C Hamano <gitster@xxxxxxxxx> writes: > > > Patrick Steinhardt <ps@xxxxxx> writes: > > > >> One topic that I miss in the cover letter is [1]. You probably forgot to > >> pick it up :) > >> > >> [1]: <20250125-b4-pks-reftable-win32-in-use-errors-v1-1-356dbc783b4f@xxxxxx> > > > > OK, I was wondering if we are going to rethink after seeing some > > comments but with blessing by j6t [*], I guess this is fine. > > > > [*]: <37c14379-68e4-4c52-a08d-b78c7797ef26@xxxxxxxx> > > Wait. This works by futzing with unlink() call that is imported via > <git-compat-util.h> which you got rid of in a separate topic, and > given the overall theme of sans-compat-util topic, I do not think > you want to include <compat/mingw.h> either directly or indirectly > (via <git-compat-util.h> as we used to), either. > > So where would that leave us...? Ah, good point. The relevant code still gets imported into the reftable library via "compat/posix.h" -> "compat/mingw/posix.h", so the fix would remain largely the same. I'll rebase the patch to fix the semantic conflicts. Patrick