Message-ID:	<B0000084465@stargate.astr.lu.lv>
From:	"Andris Pavenis" <pavenis AT lanet DOT lv>
To:	Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
Date:	Thu, 22 Apr 1999 12:31:45 +0300
MIME-Version:	1.0
Subject:	Re: [binkley AT sst DOT ncsl DOT nist DOT gov: djgpp and ld]
CC:	djgpp-workers AT delorie DOT com
References:	<199904201441 DOT KAA10263 AT envy DOT delorie DOT com>
In-reply-to:	<Pine.SUN.3.91.990422112332.15748E-100000@is>
X-mailer:	Pegasus Mail for Win32 (v3.02b14)
Reply-To:	djgpp-workers AT delorie DOT com

On 22 Apr 99, at 11:23, Eli Zaretskii wrote:

> 
> On Tue, 20 Apr 1999, DJ Delorie wrote:
> 
> > > Because ${DJDIR} works even for those who do NOT install it in /djgpp
> > > (like myself ;-).
> > 
> > For packages that can dymanically find where djgpp is installed,
> > great.  We're all set.  For the rest, which hard-code a static path, I
> > suggest we hard-code /djgpp instead of /usr/local or whatnot.
> 
> In my experience, the absolute majority of GNU packages only use
> --prefix to put it and its derivatives into Makefile's, and only in
> conjunction with "make install" or in creating absolute path names of
> programs used by the Makefile.  For these, ${DJDIR} is The Right
> Thing (tm), since it *is* dynamically resolved when the package is
> rebuilt.
> 
> For the small number of packages I saw that use --prefix to hard-code
> path names into the binaries, I also make a point to use
> --prefix=c:/djgpp, or change the program to resolve ${DJDIR} at run
> time.
> 

One of such bad examples is egcs. With egcs-1.1.X I found some workaround
(by splitting packages into 2 parts). For upcoming egcs-1.2 seems we'll have
to find another way. I'm able to build snapshots also without that but such 
hacking is not for distribution. 

Perhaps also binutils will have the same problem if we'll try to enable 
NLS support.

Andris

- Raw text -