Help us defend against VMS!

[Richard A. O'Keefe]

In article <4080 at megaron.arizona.edu>, lm at arizona.edu (Larry McVoy) writes:
> I agree with the rest of the article but this part is not completely
> true.  VMS fortran is the de facto industry standard.  Until I can have
> all the VMS extensions (and there are a lot of very useful ones) this
> argument does not hold water.  Sorry, Barry, but we can't misrepresent
> the facts.  And maybe Ultrix supports them (I don't know) but that's 
> not enough - your argument said from the PC to the super computer
> (super computer companies take the VMS extensions _very_ seriously).
> 
If this is so, it would be interesting to know why Fortran 8X does not
resemble VMS Fortran.  For example, both let you define record types,
but the syntax is very different.  IBM's FORTVS doesn't seem to have
copied much from VMS Fortran either, I've carefully checked a recent
IBM Fortran manual and couldn't find any of the features that make
VMS Fortran so sexy.  Apollo's Fortran hasn't any of the VMS extensions
that I know of.  "de facto industry standard"?  Maybe on Vaxen.

The extensions in VMS Fortran are precisely the reason why the original
point is valid:  there are so many extensions in VMS Fortran that if you
don't watch out you will write wonderful programs that are ever so much
clearer than you could have written under UNIX, but you never be able to
run them on non-DEC equipment.  (DEC's VMS C compiler is available under
Ultrix, does anyone know whether the VMS Fortran compiler is available
under Ultrix as well?)

Perhaps a survey of the features of VMS Fortran, and a list of which
super-computer companies have copied which extensions, would be a good
topic for comp.lang.fortran.

Unlike some other proprietary operating systems, VMS is not a major
nuisance for university computing.  There is a lot that is good about
it.  But it *IS* a proprietary operating system, and the compilers
available for it *ARE* proprietary compilers, and RDB *IS* a proprietary
data base.  Using UNIX gives you more freedom in selecting additional/
replacement machines.  Imagine a central VAX, with a dozen or two SUNs,
a couple of Pyramids, and maybe a Sequent or two.  Only one of them can
run VMS, but they can all run UNIX, and it isn't hard to write programs
in C or Fortran which only need recompilation to move from one to another.