Jeff King <peff@xxxxxxxx> writes: > I think this version still has the small issue that we'll _only_ surface > a generic error return in such a case, and never report EMFILE > specifically. I.e., I think we'd still want something like this on top: [snip] OK, I'll do this. This also has the advantage of not using map_loose_object_1, so I'll be able to inline it in the previous patch.
