Return-Path: <raeburn@MIT.EDU>
Received: from central-city-carrier-station.mit.edu by po11.mit.edu (8.9.2/4.7) id IAA02022; Tue, 30 Jul 2002 08:53:24 -0400 (EDT)
Received: from manawatu-mail-centre.mit.edu (MANAWATU-MAIL-CENTRE.MIT.EDU [18.7.7.71])
	by central-city-carrier-station.mit.edu (8.9.2/8.9.2) with ESMTP id IAA07632;
	Tue, 30 Jul 2002 08:52:27 -0400 (EDT)
Received: from all-in-one.mit.edu (ALL-IN-ONE.MIT.EDU [18.18.1.71])
	by manawatu-mail-centre.mit.edu (8.9.2/8.9.2) with ESMTP id IAA29807;
	Tue, 30 Jul 2002 08:52:27 -0400 (EDT)
Received: (from raeburn@localhost) by all-in-one.mit.edu (8.9.3)
	id IAA29980; Tue, 30 Jul 2002 08:52:26 -0400
To: "Hardy.Charles" <Hardy.Charles@MicroLinear.com>,
        "'sipb@mit.edu'" <sipb@mit.edu>
Subject: Re: Emacs--Maximum Buffer Size Exceeded
References: <200207300739.g6U7d9U11891@syn.hamachi.org>
From: Ken Raeburn <raeburn@MIT.EDU>
Date: Tue, 30 Jul 2002 08:52:26 -0400
In-Reply-To: <200207300739.g6U7d9U11891@syn.hamachi.org> (Bill Sommerfeld's
 message of "Tue, 30 Jul 2002 03:39:09 -0400")
Message-ID: <tx13cu1l6np.fsf@mit.edu>
Lines: 32
User-Agent: Gnus/5.090005 (Oort Gnus v0.05) Emacs/21.1 (i686-pc-linux-gnu)
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii

Bill Sommerfeld <sommerfeld@orchard.arlington.ma.us> writes:
>>   compilation):  env CC="cc -xarch=v9" ./configure
> So, I saw this, and was bemused that they'd list instructions for
> 64-bit emacs under headings which also included sun 3 and sun 386i...

What, you haven't seen the new Motorola 68000000 and x86-v9
processors?

> Anyhow, only recent sun4u systems (i.e., anything with an UltraSPARC
> II or newer) will typically be running a 64-bit-capable solaris.  I
> believe that 64-bit userland wasn't available until either solaris 2.6
> or solaris 7.

That's what I recall.

Alpha, AMD x86-64, IA64, and Iris 4D are the other 64-bit platforms
Emacs can support, at least as explicitly noted in the m/*.h files in
CVS.

Some of us on the Emacs development list were just recently talking
about making Emacs use "long long" under gcc on 32-bit platforms,
which would allow for full 32-bit addressing (and probably double the
size of the program, and slow it down).  The changes aren't completely
trivial, but probably not hard.  Before I read this email, I was just
thinking about doing some work on it....

Simply adjusting VALBITS on a 32-bit platform won't help much; I
believe it's as large as it can be while still allowing room for the
type bits and mark bit.  For a 300M file, you need at least 2 more
bits, for integers (buffer offsets) as well as for pointers.

Ken
