From jan@swi.psy.uva.nl  Tue Oct 26 10:39:57 1999
Received: from gollem.swi.psy.uva.nl (root@gollem [145.18.114.15])
	by swi.psy.uva.nl (8.9.3/8.9.3) with ESMTP id KAA11318;
	Tue, 26 Oct 1999 10:39:57 +0200 (MET DST)
Received: from localhost (localhost [[UNIX: localhost]])
	by gollem.swi.psy.uva.nl (8.9.3/8.9.3) id KAA07652;
	Tue, 26 Oct 1999 10:39:56 +0200
From: Jan Wielemaker <jan@swi.psy.uva.nl>
Organization: SWI, University of Amsterdam
To: prolog@swi.psy.uva.nl, "Charlie H. Chang" <a00chc00@nchc.gov.tw>,
        prolog@swi.psy.uva.nl
Subject: Re: compiling problem on IRIX 6.5.5f ...
Date: Tue, 26 Oct 1999 10:33:24 +0200
X-Mailer: KMail [version 1.0.21]
Content-Type: text/plain
References: <38151345.7699745B@nchc.gov.tw>
MIME-Version: 1.0
Message-Id: <99102610395601.06053@gollem>
Content-Transfer-Encoding: 8bit

On Tue, 26 Oct 1999, Charlie H. Chang wrote:
>Hi,
>I have been trying to compile SWI-Prolog 3.2.9 on my
>SGI O2 running IRIX 6.5.5f ...
>I got a 'core' while running 'configure' with
>--disable-mapped-stacks and --disable-readline.
>And the 'make' will stop at the command:
>
>   ./pl -O -o pl.prc  -b ../boot/init.pl
>with the following message:
>"*** Termination code 9  (bu21)"
>
>Has anyone ever successfully made the SWI-prolog
>run on IRIX 6.5.x ?? By the way, I use the MIPS pro
>7.2.1.3m C/C++ compiler.

These problems have been found (probably) and fixed in pl-3.3.0.
This version is not yet official, but you can pick an unofficial
copy from

	http://gollem.swi.psy.uva.nl/pub/pl/

Please note that a lot of things have changed in version 3.3.  Check
the ChangeLog.  The most important compatibility change is that
atom_chars/2 is now ISO compliant and atom_codes/2 does what the
old atom_chars/2 does.  In general, a global replace of atom_chars
by atom_codes should do the trick.  If you don't want to change
source-code, you can add:

user:goal_expansion(atom_chars(A,B), atom_codes(A,B)).

Please let me know whether these problems are solved.  Most important
changes are to the internals of the I/O, which is now based on the
open/close model with explicit streams with the Edinburgh see/tell
model build on top of this rather then the other way around.

	Regards --- Jan

