Message-Id: <3.0.6.32.19990202163547.0091d9c0@pop.netaddress.com>
X-Sender: pderbysh AT pop DOT netaddress DOT com
X-Mailer: QUALCOMM Windows Eudora Light Version 3.0.6 (32)
Date: Tue, 02 Feb 1999 16:35:47 -0500
To: djgpp AT delorie DOT com
From: Paul Derbyshire <pderbysh AT usa DOT net>
Subject: Re: performance
In-Reply-To: <199902022104.QAA08963@envy.delorie.com>
References: <3 DOT 0 DOT 6 DOT 32 DOT 19990202115636 DOT 00849100 AT pop DOT netaddress DOT com>
 <199902011747 DOT SAA11228 AT acp3bf DOT physik DOT rwth-aachen DOT de>
 <3 DOT 0 DOT 6 DOT 32 DOT 19990202115636 DOT 00849100 AT pop DOT netaddress DOT com>
Mime-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Reply-To: djgpp AT delorie DOT com

At 04:04 PM 2/2/99 -0500, you wrote:
>
>> C'mon guys. COFF is as obsolete as DJGPP 1.x. It's time to move on and use
>> an executable format worthy of the new millennium, or at least of the
>> 1990s. :-)
>
>Unfortunately, switching from COFF to anything else is harder than
>you'd expect, just because of the many places where COFF is assumed.

Hard != impossible...and sometimes hard things just have to be done like it
or not. Fact is people are chafing and straining at limitations of COFF
that weren't a problem in the decade where COFF first was used. The debug
info glitch pales in comparison to the sheer inability to get template
instantiation working using COFF, without either a) bugs or b) a kluge of
some kind or another that invariably is so horrible that it turns building
a simple C++ program into quantum physics even for people fairly
experienced at building programs.

-- 
   .*.  "Clouds are not spheres, mountains are not cones, coastlines are not
-()  <  circles, and bark is not smooth, nor does lightning travel in a
   `*'  straight line."    -------------------------------------------------
        -- B. Mandelbrot  |http://surf.to/pgd.net
_____________________ ____|________     Paul Derbyshire     pderbysh AT usa DOT net
Programmer & Humanist|ICQ: 10423848|