Date: Sun, 28 Feb 1999 15:43:38 +0200 (IST)
From: Eli Zaretskii <eliz AT is DOT elta DOT co DOT il>
X-Sender: eliz AT is
To: Paul Derbyshire <pderbysh AT usa DOT net>
cc: djgpp AT delorie DOT com
Subject: Re: I just wanted to assemble this old program under NASM....
In-Reply-To: <3.0.6.32.19990228081826.008acda0@pop.globalserve.net>
Message-ID: <Pine.SUN.3.91.990228153723.5950w-100000@is>
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Reply-To: djgpp AT delorie DOT com


On Sun, 28 Feb 1999, Paul Derbyshire wrote:

> Phase out the COFFisms and replace with
> code that is format-independent, localizing any format dependent stuff in
> one place where it is easily changed systematically.

A noble goal, but IMHO an impossible one.  Unless we want to suffer the 
kind of code bloat that comes with Windows, that is.

The FAQ explains which parts of DJGPP depend on the binary format.  I 
hope it is clear from that explanation that the enumerated parts are 
independent from each other, or cannot be easily localized.

For example, the requirement to port additional parts of Binutils (those 
which deal with other binary formats) is clearly something that is 
independent of the DJGPP core.  Without Binutils, we cannot have a 
linker, a profiler, a debugger, etc.

As another example, the COFF loader inside the stub is optimized for size 
and is therefore very hard to understand and maintain.  You don't want 
any such code in any other place where COFF images are loaded, like in 
the debugger support functions.