[fitsbits] top or bottom

[Clive Page]

In article <Pine.HPX.4.03.10005030917520.17814-100000 at galileo.csun.edu>,
Stephen Walton  <swalton at galileo.csun.edu> wrote:

>A problem which, I can't resist pointing out, is solved by Fortran-90.
[snip]
>I've believed for a long time that Fortran-90 is superior to C for most
>scientific work.

I agree with you, and the facts speak for themselves.  The only current
dispute is whether C++ or Fortran95 is better for scientific work.

As far as I can tell, astronomers in Europe have adopted Fortran90/95
to a much greater extent than those in North America.   Perhaps the early
availability of good Fortran90 compilers from various European companies
(NAG, Salford Software, EPC...) had some influence, or maybe there's some
other reason for the lack of conservatism over here?

>Just my little Fortran-90 evangelism for the day.  I suppose it also shows
>that Microsoft isn't the only force opposing innovation.

It's interesting that Microsoft put a lot of effort into producing a
Fortran90 compiler, called Fortran Powerstation, but gave up after a year
or so when they realised it wasn't up to the standard of lots of other
compilers already in widespread use.

To drag this back to FITS: one of the advantages of using Fortran90 to
call FITSIO is that you can set up a module containing interface
definitions of all the FITSIO routines you ever use, which makes sure
that most types of errors in call statements get detected at
compile-time.   Since such errors are very easy to make, and hard to
diagnose in Fortran77 (and C) programs, I find this feature alone worth the
price of a compiler.  The ability to use dynamic memory, data structures,
pointers, overloaded operators, and all the rest, comes as a bonus. C++
has a somewhat similar interface-checking feature, but not as good in my
opinion.
-- 
-- 
Clive Page,
Dept of Physics & Astronomy,               
University of Leicester.