Mailing-List: contact cygwin-help AT sourceware DOT cygnus DOT com; run by ezmlm
List-Subscribe: <mailto:cygwin-subscribe AT sources DOT redhat DOT com>
List-Archive: <http://sources.redhat.com/ml/cygwin/>
List-Post: <mailto:cygwin AT sources DOT redhat DOT com>
List-Help: <mailto:cygwin-help AT sources DOT redhat DOT com>, <http://sources.redhat.com/ml/#faqs>
Sender: cygwin-owner AT sources DOT redhat DOT com
Delivered-To: mailing list cygwin AT sources DOT redhat DOT com
Message-Id: <5.0.2.1.0.20010301190322.0244d370@mailhost>
X-Sender: paulson AT mailhost
X-Mailer: QUALCOMM Windows Eudora Version 5.0.2
Date: Thu, 01 Mar 2001 19:16:56 -0800
To: Earnie Boyd <cygwin AT cygwin DOT com>
From: John Paulson <munch AT powertv DOT com>
Subject: Re: New symlinks.
Cc: vinschen AT redhat DOT com
In-Reply-To: <3A9F0AA7.6AFCC739@yahoo.com>
References: <5 DOT 0 DOT 2 DOT 1 DOT 0 DOT 20010301173204 DOT 02448c90 AT mailhost>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"; format=flowed

At 09:51 PM 3/1/2001 -0500, you wrote:
>John Paulson wrote:
> >
> > $ cat `type -p gcc` | wc
> > cat: /usr/bin/gcc: No such file or directory
> >        0       0       0
> >
> > -- Hmmm... A user might think something broke.  I have not done a search
> >     of the archives to determine how often this asymmetry has been noted
> >     by users.  I don't recall having seen this mentioned in the year or so
> >     I've been subscribed, but then most of neurons have been fried by 
> MFC...
> >
>
>If you really want to review the negative chatter about this you should
>visit the autoconf AT gnu DOT org archives.
>
>Earnie.


Yeah, I remember that there was a mention of autoconf with respect to
the symlink proposal.  That's why I later said that this proposal should
work well "interactively".

On another note, much of the conversation has dealt with the symlink
being the leaf of a path specification.  It will be interesting to see
what happens wrt performance when all elements of a path must be checked
for symlink-ness.  I mean that in "/c/cygnus/home/foo/bar/thud" all
elements, "c", "cygnus", "home", "foo", "bar" and "thud", must be checked
for "c.lnk", "cygnus.lnk", ... "thud.lnk".

And when the partial paths _are_ followed, how will that interact with
mount points?  If there is a "/c.lnk" file and I also have a mount for
"/c", who wins?  During the expansion of the path, is the partially
de-symlinked path looked up in mount table?  (Nitpicking is my life).

(Corinna: if you've already done this all this worrying, accept my
apologies for stating the obvious.)

(Earnie: I only replied to you as I didn't want to clutter the list, but if
you think that the non-basename case has been missed in the discussion, either
you or I can post this potential gotcha).


----------------------------
John Paulson
PowerTV, Inc.
Tel: 408/777-5255
Fax: 408/777-4782
mailto:paulson AT powertv DOT com
http://www.powertv.com/


--
Want to unsubscribe from this list?
Check out: http://cygwin.com/ml/#unsubscribe-simple