Message-Id: <199809081223.OAA154476@ieva06.lanet.lv>
From: "Andris Pavenis" <pavenis AT lanet DOT lv>
To: Andris Pavenis <pavenis AT lanet DOT lv>, Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Date: Tue, 8 Sep 1998 14:28:16 +0300
MIME-Version: 1.0
Content-type: text/plain; charset=US-ASCII
Content-transfer-encoding: 7BIT
Subject: Re: autoconf v2.12 -- Bug in AC_PROG_LN_S?
CC: "Peter J. Farley III" <pjfarley AT banet DOT net>, djgpp AT delorie DOT com
References: <Pine DOT A32 DOT 3 DOT 91 DOT 980908101857 DOT 64692A-100000 AT ieva05 DOT lanet DOT lv>
In-reply-to: <Pine.SUN.3.91.980908140639.27664A-100000@is>
Precedence: bulk

Date sent:      	Tue, 8 Sep 1998 14:08:52 +0300 (IDT)
From:           	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Subject:        	Re: autoconf v2.12 -- Bug in AC_PROG_LN_S?

> 
> On Tue, 8 Sep 1998, Andris Pavenis wrote:
> 
> > Currently only way is to set needed environment variables and only 
> > after that run configure to override false settings for DJGPP.
> 
> I think the DJGPP port of `ln' should be changed so that it automatically 
> copies if the files aren't DJGPP .exe programs.  Then Autoconf's current 
> behavior will be the Right Thing.

Of course. For example building libg++-2.8.1.2 uses 'ln -s' even 
without looking it is really supported (I tried to build it it together 
with pgcc-1.1a)
 
> But even today I think what Autoconf does is more often than not 
> correct.
> 

Yes but there is enough situations where configure scripts does 
wrong things due to this problem. Then only remaining thing is to
browsing configure script and trying to fool it with some 
environment variables.

Andris.