Date: Wed, 17 Mar 1999 11:24:20 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: "Mark E." <snowball3 AT usa DOT net>
cc: djgpp-workers AT delorie DOT com
Subject: Re: symlink question
In-Reply-To: <199903161751.RAA53418@out5.ibm.net>
Message-ID: <Pine.SUN.3.91.990317112355.11125L-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp-workers AT delorie DOT com
X-Mailing-List: djgpp-workers AT delorie DOT com
X-Unsubscribes-To: listserv AT delorie DOT com
Precedence: bulk


On Tue, 16 Mar 1999, Mark E. wrote:

> Autoconf generated configure scripts uses a non-existant 
> file to check for 'ln -s' support. Since 'ln -s' works only in special 
> circumstances, it might be a good idea allow symlink() or at least 'ln -s' 
> to fail for non-existant files.

That would be firing into our own foot, because in some cases, we
*want* that test to succeed, because some packages use "ln -s" when
they install the executables, and we do want the symlinks then.  Other
packages use "ln -s" to link normal files, and that fails in DJGPP.
The problem is, at configure time you cannot know which case it is,
and, worse, some packages (like Web2c) use "ln -s" for both purposes.