Christopher Faylor wrote: > > On Wed, Sep 22, 2004 at 09:58:25AM -0400, Pierre A. Humblet wrote: > > > >Christopher Faylor wrote: > >> > >> On Tue, Sep 21, 2004 at 09:58:40PM -0400, Pierre A. Humblet wrote: > >> >It's a safe time to take care of a few nits... > >> > > >> >2004-09-22 Pierre Humblet <[EMAIL PROTECTED]> > >> > > >> > * path.cc (normalize_win32_path): Only look for : in second position. > >> > Avoid infinite loop with names starting in double dots. > >> > (mount_info::conv_to_win32_path): Do not worry about a trailing dot. > >> > >> Why not worry about a trailing dot? Is it handled somewhere else? The > >> intent is to make the inode of /foo/. == /foo . > > > >Yes, that's handled in the appropriate normalize_xx_path function, > >together with xx/./yy , xx/../yy, and friends. > > I'm just dotting i's but I notice that dir.cc calls hash_path_name with > an argument of ".". That still works?
It's not affected by the change, so it works as well as before. That particular piece of code is only called when the dir is ".." but the parent directory can't be identified, so the result is questionable anyway... I think it can happen when reading a directory like 'c:some_dir' Pierre