Mailing-List: contact cygwin-help AT cygwin DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT cygwin DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT cygwin DOT com>
List-Help: <mailto:cygwin-help AT cygwin DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT cygwin DOT com
Mail-Followup-To: cygwin AT cygwin DOT com
Delivered-To: mailing list cygwin AT cygwin DOT com
Message-Id: <3.0.5.32.20040327120522.007febd0@incoming.verizon.net>
X-Sender: vze1u1tg AT incoming DOT verizon DOT net
Date: Sat, 27 Mar 2004 12:05:22 -0500
To: cygwin AT cygwin DOT com
From: "Pierre A. Humblet" <Pierre DOT Humblet AT ieee DOT org>
Subject: Re: lstat on FAT - Was: Problem with find on FAT drives
In-Reply-To: <20040326200315.GA6997@redhat.com>
References: <20040325023929 DOT GA335381 AT hpn5170x> <Pine DOT LNX DOT 4 DOT 44 DOT 0403231316380 DOT 32093-100000 AT ashi DOT FootPrints DOT net> <4060B627 DOT 8000400 AT scytek DOT de> <20040324155332 DOT GF17229 AT cygbert DOT vinschen DOT de> <4061E702 DOT 1040705 AT scytek DOT de> <20040324205644 DOT GA4580 AT redhat DOT com> <20040324213057 DOT GA296803 AT Worldnet> <20040324214239 DOT GA6079 AT redhat DOT com> <20040325023929 DOT GA335381 AT hpn5170x>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"

At 03:03 PM 3/26/2004 -0500, Christopher Faylor wrote:
>[just to provide a non-flip answer to this subject]
>On Wed, Mar 24, 2004 at 09:39:29PM -0500, Pierre A. Humblet wrote:
>>On Wed, Mar 24, 2004 at 04:42:39PM -0500, Christopher Faylor wrote:
>>I wonder if 
>>      char *p = strrchr (src, '\0');
>>      /* Detect if the user was looking for a directory.  We have to
strip the
>>should be inside the symlink loop or outside. I guess that depends if
>>symlink contents ending with / are special (on Sun the final / is
stripped in
>>symlinks, dunno about other Unix flavors).
>
>I really hated putting that in there to begin with (and it should be a
>'strchr' anyway) but it was required because Windows allows you to say
>/foo/bar/ and even /foo/bar/. even when bar isn't a directory.  I don't
>believe that the code would work right if that wasn't there.  It would
>allow a symlink to /foo/bar/. to work when it shouldn't.

OK, but there are two other related issues:

On Sun
1-everest$ ln -s /etc/passwd/ pw
1-everest$ tail -1 pw
+:x:::::/var/adm/local/nologinsh
1-everest$ ls -l pw
lrwxrwxrwx   1 humblet  cm            11 Mar 27 17:49 pw -> /etc/passwd
                 (note the final / is gone)
On Cygwin:
~: ln -s /etc/passwd/ pw
~: tail -1 pw
tail: pw: No such device or address   (because the test is in the symlink
loop)
~: ls -l pw
lrwxrwxrwx    1 pierre   all           121 Mar 27 11:47 pw -> /etc/passwd/

And also
On Sun:
1-everest$ ln -s /etc et
1-everest$ ls -ld et
lrwxrwxrwx   1 humblet  cm             4 Mar 27 17:52 et -> /etc
1-everest$ ls -ld et/
lrwxrwxrwx   1 humblet  cm             4 Mar 27 17:52 et/ -> /etc
  Note that the results are identical

On Cygwin:
~: ln -s /etc et
~: /bin/ls -ld et 
lrwxrwxrwx    1 pierre   all           106 Mar 27 11:54 et -> /etc
~: /bin/ls -ld et/
drwxr-xr-x   11 pierre   all             0 Oct 12  2001 et/
  Here they differ.

Can someone check how Linux behaves in these cases?

>>Also normalize_posix_path strips the final /, except when it calls 
>>normalize_win32_path. That makes the code go through extra hoops
>>when resolving c:/the/symlink/, it looks for c:/the/symlink/.lnk
>
>I suppose that should be fixed.

Will do, after 1.5.10 is released. Don't want to break it while
fixing weird corner cases.

Pierre


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Problem reports:       http://cygwin.com/problems.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/