[log in to unmask] wrote:

> > > What are they going to do with
> > > such a monstrous, bloated Fortran - write Interactive
> > > games, operating systems, telephony software ???????


Then Roger Glover wrote:

> > Or maybe write 3D stack depth migration modeling of seismic
> > data like at least two oil companies I know of are doing
> > with their C++ code...


Then Pierre Hugonnet wrote:

> There's one point that I can talk about: I'm working in an
> oil company and I claim that C(++ or --) is definitly *not useful*
> to write seismic data processing code, at least to write the
> scientific part of the code!

You have me at a disadvantage.  In the immortal words of Tom Cruise,
"I could tell you, but then I'd have to shoot you."  I cannot be
specific about either of the two companies, but I can say that in
both cases effective parallelization of floating-point-intensive
C++ 3DSDM code was a key factor in a competitive procurement.
Because I cannot be more specific, I cannot prove these assertions
to you; I will withdraw them if you insist.

I can report specifics of a third case in the oil industry, but it
seems to have more to do with future direction than current practice.
Last year, a colleague and I taught a seven-week series of classes
for ARAMCO in Saudi Arabia, of which 4 were C++, 3 were OOP, and 1
was Fortran 90.  There was not an empty seat in any of the C++ or OOP
classes, but the F90 class was half-empty.  This year, my colleague
returned alone to teach a four-week series:  2 C++, 1 OOP.  In my
classes most of the students were involved in seismic processing,
with maybe one out of ten doing reservoir simulation.

This is not *MY* choice for how things should turn out, but it is the
fact.  What does it prove?  Only that there is a great deal of
interest in C++/OOP at one of world's largest oil companies.  Would
they be interested in Fortran with objects?  I don't know.


> Here some people use Fortran
> (most of them 77), and some people use C. But when I look at the
> code of the latter ones, I'm a little bit puzzled...
> Why do they use C(++) ?:

People often use C++ for utterly different reasons than C.  When
used to its best advantage C++ is a completely different language
that happens to contain most of C as a subset.


> - They've been taught that Fortran writers escaped from Jurassic Park,

Prejudice is no uglier on their side of the fence than on ours.
"You mean those BARBARIANS think WE are DINOSAURS?  WHAT NERVE!"


>   but all of them write either unreadable C code, or Fortran like C code.

If you are fluent in Fortran but not in C, *ALL* C code is, by
definition, either "Fortran-like" or "unreadable."


> - They need X11/motif capabilities: I agree with them that Fortran
>   lacks graphical interfaces.

There are graphics and GUI libraries for Fortran (check the Fortran
Market), but they are generally proprietary, extremely hardware- or
OS-dependant, and/or less than state-of-the-art.  I would be more
than happy for someone to prove me wrong.


> - They want to build OO industrial applications (don't laugh please):
>   I'm afraid that software engineering is a profession, and that most of
>   scientists (including me) are simply not able to produce good OO code.
>   the point is not only the knowledge of the syntax...

Lab technician is a profession too, but that does not stop most
experimental chemists from being excellent lab techicians.  Using
object technology is a skill on par with being able to work safely
and effectively in a chemistry lab, but *not* on par with most true
scientific and engineering work.  I have no doubt that if you thought
it was truly important to your work you could pick up the basics in
just a few weeks and be thoroughly skilled within a year.


-------- Cray Research --------- Roger Glover
-- A Silicon Graphics Company -- http://home.cray.com/~glover


%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%