Like it or not, COBOL is still used by business.

btw, academia is now busy playing "Nintendo Science" with shrink wrapped
toys  translated from netlib code. Who knows, perhaps half of those hits
are by the copy wizards of Matlab et al!?

Jim

I heard recently (from a member of the Fotran committee) that
it's difficult to get any statistics, but the total usage of Fortran
may now be at an all time high.  As a percentage of the total
computing language market, Fortran has shrunk (and may
be continuing to shrink).  But that's because all of the new
application directions (communications, entertainment,
commerce, etc.) are written in other languages - and they're
bigger industries than any Fortran has ever tackled.

So, the use of Fortran could actually have been monotonically
growing all the while, and it just appears to be shrinking because
other languages have grow so much faster.  How would such
a statistic be verified (or disproven)?  Even people that use
Fortran heavily might feel that it's unstylish and resist participating
in a survey about it.

Working from a different direction, consider the issue of vendor licenses.
HP has a large number of current support contracts for Fortran licenses (I
suspect the exact number might be considered business competitive
information, so I won't disclose it).  But that number tells us little about
the number of current users who write Fortran code.

How many of these licenses are on machines where Fortran is only used to
compile programs provided by third party applications?  How many are on
single user work stations?  How many are on multiuser servers with active
program development?  And how many are on machines where it was part of the
package deal and Fortran has never been used on that machine?
You might investigate a scientific sampling throughly, but that still
could provide a misleading picture, depending on how you phrased
your questions.

From informal conversations with customers, their interests span the gamut,
from those who want all the latest F95 features, plus OpenMP and other
language extensions, to those who just want their old Fortran77 (of various
dialects and extensions) to just keep working.  If I wanted to bias the
results of any survey, even with carefully designed questions, I only
need to preselect who the questions will be sent to.  Even without
a malicious intent to bias the results, the way I pick my sample will
have a strong impact on the answers that are received.

To give some examples of different user communities:
(1) Supercomputer users vs Workstation users
(note that I know that supercomputer users also use workstations,
 I mean by Workstation users those who can get all their work
 done on their personal workstation)
The former will have an interest in parallel programming extensions.
The latter won't.

(2) Research labs vs production shops
The former is more likely to have those interested in Fortran90 to aid in
their application development, while the latter is more focussed on whatever
always worked, cause they 'have a job to get done now'.

(3) Different branches of science and engineering
Some are more inclined to shift to Fortran90 than others, perhaps
because of the actions of opinion leaders within their discipline
5 years ago.

So broad surveys of Fortran users attitudes are perhaps less useful than
targeted surveys of particlar types of users.

Your mileage may vary...

And Fortran programmers don't read c.l.f. In their minds, c.l.f. is for
Fortran experts. A month ago, I offered to give a talk on Fortran
optimization here. With about no advertising, 25 people attended,
none of them knew about c.l.f., C-Fortran language wars or even that I, in the
same building, was active writing utilities and discussing about the standard.
And none of them would call him/herself a programmer.

Michel

The fact that the CS department doesn't want to teach Fortran should not be
a big problem.  In fact, it can be an good thing because they would teach it
all wrong.  You can turn the situation to your advantage, though, by
offering an UG class in Fortran in your own department.

At UCSD, the AMES (Applied Mechanics and Engineering Science) Department did
just that, with excellent results.  In fact, it was a class that the faculty
really enjoyed teaching, because it required no preparation - any professor
could teach the class with his eyes shut, practically on autopilot.

One benefit of having your own class is that you can teach the students how
to quickly and easily put together programs that work, and solve real
problems.  We all know that the real agenda of the CS department everywhere
is to bury all software production underneath a mountain of pointless and
endless design specifications that turn even a simple issue into a
Kafkaesque odyssey.  Their primary intention to make computers inaccessible
to the average user in any direct sense, thus forcing all users to pay CS
"experts" and "specialists" to do the work instead.

snip

01100001100001100001111100011
11110011111000000111100111001

And as others have pointed out, most Fortran gets written by engineers or
scientists.   Around here that still happens, with only a small defection
to more fashionable languages such as C++ or Java.

I've put on a Fortran77-to-Fortran90 conversion course a few times, and
got 25 - 30 takers each time.  But I doubt if any of them knew of the
existence of comp.lang.fortran or the fortran90 mailing list either
(before I told them).   People like that don't get counted in statistics of
language use.