Newsgroups: comp.sys.ibm.pc.misc
Path: utzoo!utgpu!news-server.csri.toronto.edu!helios.physics.utoronto.ca!alchemy.chem.utoronto.ca!mroussel
From: mroussel@alchemy.chem.utoronto.ca (Marc Roussel)
Subject: Re: When will the 8088 die?
Message-ID: <1990Dec4.160730.15617@alchemy.chem.utoronto.ca>
Organization: Department of Chemistry, University of Toronto
References: <90335.202651F0O@psuvm.psu.edu> <3360005@hpsgwp.sgp.hp.com> <1990Dec4.014539.13773@magnus.ircc.ohio-state.edu>
Date: Tue, 4 Dec 90 16:07:30 GMT

In article <1990Dec4.014539.13773@magnus.ircc.ohio-state.edu>
smsmith@hpuxa.ircc.ohio-state.edu (Stephen M. Smith) writes:
>My point was that software becomes *more* expensive when the
>programmer is forced to right it for 8088, 286, 386, monochrome,
>CGA, EGA, VGA, etc., etc....  

     Most commercial software is written in a high-level language.
Since 8088 compilers already exist, it is then just a question of
running the same program through different compilers to have versions
that will run reasonably efficiently on a given architecture.  Sure you
have to tweek the program on each architecture, sure a "real" 386
version is more than a mere recompile, but it need not involve a
complete rewrite.
     As to your point about the different graphics formats, most
professional programmers have some kind of library of graphics
primitives for each of the popular graphics formats, so it's not even a
question of recompiling, just relinking.
     I think you're making too much of a big deal of this backward
compatibility thing.  You can have backward compatibility of your chips
without compromising the efficiency of the design of the silicon, so why
not?  I can just see it now... "You can have this super-duper new chip
in your machine but you'll have to throw away all your software."

				Marc R. Roussel
                                mroussel@alchemy.chem.utoronto.ca
