On Tue, Nov 28, 2017 at 5:04 PM, Jonathan Nieder <jrnie...@gmail.com> wrote: > (+cc: bug-gnulib@, since they share this code)
> If I understand correctly, this treats DT_UNKNOWN like DT_DIR when stat > indicates that it is a directory. The patch checks if an entry is really a dir if type is DT_UNKNOWN. > > What should happen in the DT_LNK case? Should the same logic trip for > it as well so we can distinguish between a symlink to a directory and > other symlinks? You are right. i'll add a test and post another patch. regards, Dmitry