Path: utzoo!yunexus!geac!dmntor!steve
From: steve@dmntor.UUCP (Steve Pozgaj)
Newsgroups: comp.os.misc
Subject: Re: QNX anyone?
Summary: Caveat Emptor
Message-ID: <1500@dmntor.UUCP>
Date: 27 Apr 89 17:04:52 GMT
Article-I.D.: dmntor.1500
References: <8904252052.AA05490@decwrl.dec.com>
Reply-To: steve@dmntor.UUCP (Steve Pozgaj)
Organization: Digital Media Networks Inc., Toronto, Canada
Lines: 28

We have been forced to use QNX, by virtue of inheriting applications.
Without going into many gorey details, which I'd be happy to discuss
individually, we are converting to UNIX [on SUNs, in our case] ASAP.

I have consistently maintained that the only similarity between QNX
and UNIX is they both contain the letters "N" and "X".  Seriously.
QNX is probably a wonderful system for running factory automation,
process control, or tasks of that "almost-realtime" ilk.  However,
for software development, my vote, seconded with a tremendous roar
by my software development staff, is a resounding "NO WAY!".

Why?  Again, in a nutshell, one of the most difficult undertakings in
software development is to write code on an inherently unreliable
platform.  QNX is that, in my opinion.  When one cannot rely on the
operating system's fundamental commands to execute properly, one gets
chaos ... and many wasted days of expensive programmer effort.  The
QNX shell is the root of this unreliability; unfortunately, it spreads
all the way through many of the underlying commands.

This is not to say QNX is without merit, as suggested above.  However, the
question the original poster posed was QNX's use in *developing* software.
The answer, again, must be "no".  Use the real thing; that's what it was
designed for, and does it ever show!

P.S. The *people* behind QNX are tremendously helpful.  They always tried
to either solve the myriad problems we uncovered (and, unfortunately,
continue to uncover), or, at least, put us in touch with people who know.
I only wish their product were as good as their intention.