From: bdc AT martigny DOT ai DOT mit DOT edu (Brian D. Carlstrom)
Subject: RE: Bug in find?
19 May 1998 02:21:05 -0700
Message-ID: <199805181659.AA062110760.cygnus.gnu-win32@martigny.ai.mit.edu>
References: <c=GB%a=TMAILUK%p=DCNET%l=EXCHANGE2-980518141128Z-11280 AT smtp DOT datcon DOT co DOT uk>
Reply-To: "Brian D. Carlstrom" <bdc AT ai DOT mit DOT edu>
To: Edward Avis <EPA AT datcon DOT co DOT uk>
Cc: "'Brian D. Carlstrom'" <bdc AT ai DOT mit DOT edu>,
        "'gnu-win32 AT cygnus DOT com'"
	 <gnu-win32 AT cygnus DOT com>

Edward Avis writes:
 > >I made the observation that find was not descending into directories
 > >that had a repeated path segment such as findlosage/quux/foo/foo or
 > >findlosage/quux/foo/quux. This was running the find starting in the
 > >findlosage directory.
 > 
 > This is because of the simulated inode numbers.  These are calculated
 > based on the directory name, and hence foo/bar has the same inode number
 > as foo/bar/bar.  On Unix, it makes sense for find to skip a directory if
 > it has already seen that inode - but it gets confused on Cygwin32.
 > 
 > AFAIK, this problem is not as bad on B19, because the whole path is used
 > to seed the inode number, not the leafname.  So if you're running B18,
 > I'd suggest moving to the newest version (and coolview, etc.)

I'm sorry I forgot to mention that this is a problem with B18, B19, and
B19.1. I actually did an 'ls -i' to make sure the inode numbers were
different and they seemed to be...I also tried the -noleaf options to
find just for voodoo sake.

-bri
-
For help on using this list (especially unsubscribing), send a message to
"gnu-win32-request AT cygnus DOT com" with one line of text: "help".