From: sandmann AT clio DOT rice DOT edu (Charles Sandmann)
Message-Id: <9606221644.AA11100@clio.rice.edu>
Subject: Re: linking speedup ?
To: k3040e4 AT c210 DOT edvz DOT uni-linz DOT ac DOT at (Oberhumer Markus)
Date: Sat, 22 Jun 1996 11:44:14 -0600 (CDT)
Cc: djgpp-workers AT delorie DOT com
In-Reply-To: <199606221644.SAA01348@c210.edvz.uni-linz.ac.at> from "Oberhumer Markus" at Jun 22, 96 06:44:40 pm
Content-Type: text
Content-Length: 672       

> After a look at the specs I suspect this is because gcc calls stubify
> after ld created the COFF image. Is this true ?

This is correct, yes, and does contribute to the slowness.

> If yes, would it be much work to add the functionality
> of stubify to ld if the output file ends in '.exe',
> thereby saving a write and read of the image ?

This is what DJ calls "teaching BFD about the stub".  Once the GNU routines
are fixed to know about the stub, all sorts of good things happen, like
size/strip/gdb can handled stubbed .EXE images, ld could write the stub 
directly, etc.  So far noone has had the time and energy to make these
changes and submit them to the FSF.