Path: utzoo!utgpu!news-server.csri.toronto.edu!cs.utexas.edu!wuarchive!zaphod.mps.ohio-state.edu!rpi!batcomputer!cornell!uw-beaver!zephyr.ens.tek.com!tektronix!percy!m2xenix!puddle!p25.f506.n106.z1.fidonet.org!Jon.Guthrie
From: Jon.Guthrie@p25.f506.n106.z1.fidonet.org (Jon Guthrie)
Newsgroups: comp.lang.modula2
Subject: Re: Oh My! Modula-2 Book
Message-ID: <3114.27EAAB68@puddle.fidonet.org>
Date: 20 Mar 91 20:23:48 GMT
Sender: ufgate@puddle.fidonet.org (newsout1.26)
Organization: FidoNet node 1:106/506.25 - Fulcrum's Edge, Spring TX
Lines: 49


 On a message of 18-Mar-91, Aubrey McIntosh Said:

> But I can buy the source to the libraries, and if they are well
> written, port them into the alternate compiler's environment.

Yes, that IS possible.  Now try to ship production code with those
ported libraries imbedded in them.  If you ever get over the legal
difficulties, get back to me and I'll say 'I told you so.'

> If there was a standard, and "Primo Donya Software" released a
> non-compliant version on "Spiffo Cpu" before anyone else, then we'd
> be right where we are now without a standard.

No, that isn't necessarily true.  (I think you're probably trying to
draw an analogy with Turbo Pascal.)  If there WERE a standard, there
would be strong advantages in following it.  (Which is why most
C compiler vendors have already moved to ANSI C and are moving to C++
despite other alternatives that are every bit as good.)

> If this is a REAL nuisance in your mind, then invert that to a
> potential income advantage:  Write a correct and portable library,
> and methodically place it on each available platform/vendor
> position.  Either you should get rich, or you shouldn't be bitchin.

You speak as if that were child's play to get compiler vendors to
accept outside code.  You also speak as if most people out there are
using Modula-2.  Might I remind you that an awful lot more work is
done in C (which has had a well-defined I/O and support library almost
since it's creation) than in M-2 and the lack of a library standard is
one of the primary reasons.

The REASON it's a difficulty because you are forced to learn another
way of doing things every time you get a new compiler.  Logitech
expertise doesn't translate into Stonybrook which doesn't translate
into TopSpeed which doesn't translate into FST.  You can't say "I'm a
Modula-2 programmer" without adding some kind of statement about which
compiler you're using.

While I can appreciate Randy Bush's point about there being no clearly
superior way to define an I/O library, I must insist that C is an
example that shows how powerful even a mediocre (and often actively
bad) library I/O standard can be when it comes to portability.



--  
uucp: uunet!m2xenix!puddle!106!506.25!Jon.Guthrie
Internet: Jon.Guthrie@p25.f506.n106.z1.fidonet.org