"Shawn O. Pearce" <spearce@xxxxxxxxxxx> writes: > Nicolas Pitre <nico@xxxxxxx> wrote: >> First, I truly believe we should have a 64-bit pack index and fewer >> larger packs than many small packs. > > I'll buy that. ;-) > >> Which leaves us with the actual pack index lookup. At that point the >> cost of finding an existing object and finding that a given object >> doesn't exist is about the same thing, isn't it? > > Here's the rub: in the missing object case we didn't find it > in the pack index, but it could be loose. That's one failed > syscall per object if the object isn't loose. If the object > isn't loose, it could be that it was *just* removed by a > running prune-packed, and the packfile that it was moved > to was created after we scanned for packfiles, so time to > rescan... If that is the only reason we have these reprepare_packed_git() sprinkled all over in sha1_file.c (by 637cdd9d), perhaps we should rethink that. Is there a cheap way to trigger these rescanning only when a prune-packed is in progress, I wonder... - To unsubscribe from this list: send the line "unsubscribe git" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html