Help us defend against VMS!

[Richard A. O'Keefe]

In article <4125 at megaron.arizona.edu>, lm at arizona.edu (Larry McVoy) writes:
> And how many cycles do you think apollos spend chewing on fortran?  Or 
> ibm's?
Lots.  When Apollo brought out their first machines, they were
AEGIS/Pascal/Fortran machines.  And yes, people do buy IBM mainframes
and run Fortran on them.  (Stop and think about 3090/VF.)
I have heard of some large sales of PR1ME superminis to outfits
that wanted to run particular (Fortran-coded) packages on them.

If you're interested in Fortran cycles, ask who uses SPSS, who uses
PAFEC, who uses GENSTAT, who uses ... 

> To beat a dead horse: would you buy
> an apollo to run fortran?  Or a sun for that matter?  Or an ibm?

I wouldn't buy an Apollo or a /370 myself, but if I had some Fortran
to crunch, and the money to pay for it, I would certainly consider a
Sequent.  (Their Fortran doesn't look like VMS Fortran either.)
If my business involved trying to write reasonably portable packages,
VMS is the *last* thing I would use, precisely because it is so
different from other Fortrans.  And I would prefer a Sun to a VAX
because I'd get better performance/price from it.

> I think that you should ask fortran hackers what sort of fortran they
> want, not Unix/C hackers.

If ability to port existing code is the issue, what matters is what sort
of Fortran they've _got_, not what they _want_.  I'd suggest a survey of
comp.lang.fortran, except that's likely to be biassed in favour of UNIX.
Somewhere there must surely be published figures on relative Fortran usage
of various machines.

My point that Fortran 8X does not strongly resemble VMS Fortran remains
unchallenged.  I find it difficult to reconcile that with the claim that
VMS Fortran is the "de facto industry standard".