Date: Thu, 27 Nov 1997 14:30:28 +0100 (MET)
From: Hans-Bernhard Broeker <broeker AT physik DOT rwth-aachen DOT de>
To: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
cc: DJ Delorie <dj AT delorie DOT com>,
        djgpp workers list <djgpp-workers AT delorie DOT com>
Subject: Re: alpha-971114: next round of Makefile changes
In-Reply-To: <Pine.SUN.3.91.971127132843.5743I-100000@is>
Message-ID: <Pine.LNX.3.93.971127142445.16341A-100000@acp3bf>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk

On Thu, 27 Nov 1997, Eli Zaretskii wrote:
> On Wed, 26 Nov 1997, Hans-Bernhard Broeker wrote:
> > * Unzipping the sources on Linux (with '-a' to get rid of CR's) exposes
> > some filename problems, mainly the .s <--> .S issue. Real fix would of
> > course be to zip them up as '.S' right away. Some problems will not be
> > fixed that way though.
 
> Which files are those, and why a correct file name doesn't fix that?

The main problem is in two places:

1) In src/libc/crt0, the files 'crt0.s' and 'gcrt0.s' have to be renamed,
otherwise 'make' won't find them (it explicitly names them in rules, so
the usual trick with the pattern rule %.o: %.s doesn't work here).  But
once they are renamed, 'gcrt0.S' still won't compile, as it #includes
'crt0.s' instead of 'crt0.S'. 

2) 'make clean' in src/libc/stubs removes '*.S', but here, the files
generated by 'mkstubs.exe' are actually called '*.s', even on Linux. 

The patch set fixes 2), but for 1), I only offer a shell script to
do the necessary changes.

Hans-Bernhard Broeker (broeker AT physik DOT rwth-aachen DOT de)  Even if all the
snow were burnt, ashes would remain.