On Wed, Mar 10, 2021 at 4:04 AM Ævar Arnfjörð Bjarmason <avarab@xxxxxxxxx> wrote: > > [I couldn't find a more relevant thing to reply to] > > On Thu, Jul 30 2020, brian m. carlson wrote: > > > [...] > > B.t.w. thanks again for all your work on SHA-1 -> SHA-256. > > I found a missing spot that wasn't trivial to fix, so sending an E-Mail: > > In git-filter-branch.sh we have: > > _x40='[0-9a-f][0-9a-f][0-9a-f][0-9a-f][0-9a-f]' > _x40="$_x40$_x40$_x40$_x40$_x40$_x40$_x40$_x40" > > Then later we have a case condition based on matching a SHA-1: > > $_x40) > echo "Ref '$ref' was rewritten" > if ! git update-ref -m "filter-branch: rewrite" \ > > Just deleting that case arm has filter-branch tests passing, so whatever > it's meant to do it has zero coverage, which explains why it hasn't > broken with our tests. > > I didn't have time to dig, so sending off this E-Mail instead. Hmm, perhaps we should revive the "separate non-core tools out of git.git" thread -- https://lore.kernel.org/git/a784a61e-1320-be1e-9dfb-d533a01827ec@xxxxxxxxx/. Honestly not sure it's worth bothering fixing filter-branch at this point, given the strong WARNINGs we added to it. If anything, I'd just add another warning to the pile for this issue. If folks really still want/need filter-branch once SHA-256 is the only option, I'm sure they'll be motivated to make the necessary fixes (or just use/fix filter-repo's filter-branch-ish).
