From: elf AT netcom DOT com (Marc Singer)
Message-Id: <199603262001.MAA01655@netcom23.netcom.com>
Subject: Re: djgpp compiling real mode?
To: A DOT APPLEYARD AT fs2 DOT mt DOT umist DOT ac DOT uk (A.Appleyard)
Date: Tue, 26 Mar 1996 12:00:58 -0800 (PST)
In-Reply-To: <11C66561CAA@fs2.mt.umist.ac.uk> from "A.Appleyard" at Mar 26, 96 10:20:23 am
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Content-Length: 1404
Sender: elf AT netcom DOT com

>  I wrote suggesting that djgpp should be able to compile small
> real-mode functions and programs (e.g. as interrupt
> handlers). Someone replied that the resulting (need to cope with
> segment-and-offset type addresses in 32bit mode) would cause too
> much complication. This complication would be avoided if the rule
> said that any djgpp real-mode compilations must all fit (including
> all working space) into 64K bytes and thus all in one segment and no
> heap / malloc / new / delete / etc, no recursion, all array bounds
> to be numeric. This would be enough to write interrupt handlers and
> similar without the user having to clutter his computer with another
> C compiler to write real-mode odds and ends.

Seems to me (IMHO, I reckin, and all of that) that the issue is not
with DJGPP, but with GCC.  While I suppose you could modify GCC to
target a real-mode 80x86, it would be WAY more work than anyone wants
to tackle.  My personal desire is to help shoo DOS out the doggie door
as fast as possible so we can get on with writing software.  That is
opposed to thrashing about, finding `work-arounds' for all of the
inept design decisions that went into the beast.  Note that thats is
me speaking for myself and in no way trying to influence the opinions
of any other person or animal.

-- 
Marc Singer      |                     If you give it water, it'll turn green.
elf AT netcom DOT com   |