Date: Thu, 8 May 1997 16:46:21 +0200 (MET DST)
From: Mark Habersack <grendel AT hoth DOT amu DOT edu DOT pl>
Reply-To: grendel AT hoth DOT amu DOT edu DOT pl
To: Chris Matrakidis <cmatraki AT eleceng DOT ucl DOT ac DOT uk>
cc: djgpp-workers AT delorie DOT com, djgpp AT delorie DOT com
Subject: Re: PE executable format
In-Reply-To: <707.9705081442@baldrick>
Message-ID: <Pine.BSI.3.96.970508164420.4125A-100000@hoth.amu.edu.pl>
Organization: PPP (Pesticide Powered Pumpkins)
MIME-Version: 1.0
Content-Type: TEXT/PLAIN; charset=US-ASCII
Precedence: bulk

On Thu, 8 May 1997, Chris Matrakidis wrote:

> > For PE we'd have to write code to output M$ COFF format, a linker for it and
> > a loader.
> 
> As of binutils 2.7 the PE format is supported in the standard distribution.
> And now binutils 2.8 are out, where this support should be more stable.
Ah... didn't know that, sorry!
 
> > With ELF we'd only have to add a new target to DJGPP and write a loader.
> > Adding new target is just a matter of enabling the new format support in BFD
> > and recompiling the compiler.
> 
> I agree that ELF support is easy as a target for the linker. However, the
> support that is necessary in the library and the djgpp utilities (like symify)
> is harder.
Hmm... wouldn't it be equally hard for PE (it is COFF, but not quite compliant
with the original specs)

> My point was that in the MSDOS/Windows world there are many tools that support
> PE executables. Compatibility with them can only benefit DJGPP.
True, but I'm also thinking about the benefits of a format itself. ELF is much
more modern and better designed than COFF in general and PE in particular. If
changing/adding the output format, why not to use the best one?