Date: Tue, 21 May 1996 08:06:58 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
To: C D Radcliffe <men5cdr AT sun DOT leeds DOT ac DOT uk>
Cc: djgpp AT delorie DOT com
Subject: Re: Help required with DPMI Conflict
In-Reply-To: <1996May20.195533.28424@leeds.ac.uk>
Message-Id: <Pine.SUN.3.91.960521080045.13486C@is>
Mime-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII


On Mon, 20 May 1996, C D Radcliffe wrote:

> On upgrading to DJGPP V2.0 and recompiling GNUPLOT I find a problem when
> I spawn a shell. From within GNUPLOT I would normally spawn a shell and
> run ghostscript (V3.1.2 Watcom DOS4/GW compiled version). If I do that
> with the V2.0 compiled Gnuplot my machine hangs/crashes. 

If Ghostscript is compiled as 16-bit DPMI client, this won't work (16-bit
and 32-bit DPMI clients cannot be mixed, this is a ``feature'' of DPMI 0.9
spec).  If Ghostscript is compiled as a 32-bit program, then a possible
cause might be that you run DJGPP without any memory manager (such as
EMM386 or QEMM); in that case CWSDPMI reserves all the extended memory to
itself, so Ghostscript doesn't find any extended memory when it is
spawned.  I don't know if there is another conflict between DOS/4GW and
CWSDPMI, but there might be.  Did you try to do the same under a different
DPMI host (e.g. in Win3.x DOS box)? 

The best way would be to compile Ghostscript under DJGPP v2.