Han-Wen Nienhuys <hanwen@xxxxxxxxxx> writes: > On Tue, Feb 1, 2022 at 3:02 AM Junio C Hamano <gitster@xxxxxxxxx> wrote: >> >> "Han-Wen Nienhuys via GitGitGadget" <gitgitgadget@xxxxxxxxx> writes: >> >> > From: Han-Wen Nienhuys <hanwen@xxxxxxxxxx> >> > >> > As discussed in >> > https://lore.kernel.org/git/xmqqbkzrkevo.fsf@gitster.g/ , we don't >> > want to treat the sign bit specially, so make all flags in refs.h >> > unsigned. >> >> It may be true that we do not want to, but are we sure there is no >> code that already uses some cleverness to make the topmost bit >> special by e.g. doing ugly things like "if (flags < 0)" ;-)? > > In refs.h, we don't have any bitmasks that accept 32 different bit > values, so it's hard to see how we would stumble into a case like > that. > >> I haven't had energy to vet the entire codebase for all flag words >> and that is why I've never produced a patch like this one myself. >> >> Thanks for carefully vetting after creating this (presumably >> mechanically prepared) patch---producing is the easy part, but >> making sure it does not break unexpectedly is much harder. > > I went over the changes once more, uniformizing the naming. I didn't > see any shenanigans. Thanks again.
