Path: utzoo!censor!geac!torsqnt!news-server.csri.toronto.edu!cs.utexas.edu!swrinde!zaphod.mps.ohio-state.edu!think.com!mintaka!yale!venus!tang
From: tang@venus.ycc.yale.edu
Newsgroups: comp.lang.fortran
Subject: Re: Fortran vs. C for numerical work
Message-ID: <671.275e3c4a@venus.ycc.yale.edu>
Date: 6 Dec 90 17:40:41 GMT
Organization: Yale Computer Center (YCC)
Lines: 40

Marc Roussel writes
>     I am in graduate school in a very active research group.  Almost all
>of us use Fortran in this department, except for system administration
>and X-window programming.  I think that your comments underestimate the
>strength of Fortran.  Physicists, chemists and other natural scientists have
>different needs and interests than computer scientists.  Fortran suits
>many of these needs and will continue to be learned by natural
>scientists although I concede that computer scientists can now safely
>ignore it.  Fortran has returned to the niche for which it was originally
>intended.  Similarly, Cobol is no longer used to write OS's, but will
>probably continue to be used in business programming as it provides a
>natural interface in this environment.  (I know a lot less about Cobol
>than I do about Fortran usage... If Cobol is really dying, as the CS
>types are constantly assuring me, then you may ignore the last sentence
>of this paragraph.)

Absolutely!  

However, since languages are part of CS, and if there were enough CSists 
around to convince people who have not made up their mind yet, we might 
see a demise of Fortran.  

There is a growing trend towards using more elaborate data structures.
For example, discrete-particle simulations are frequently used in numerical
statistical mechanical systems, evolution of solid structures under random
growth environment (diffusion is really a random process).  In just these
two cases, sorting, and random access of elements are the likely to be the 
most time-consumming tasks, Fortran may not be the appropriate language
(correct me if I am wrong).  

IMHO, these capabilities should be added to the
language (Fortran 90 may have them, I do not know much about F90) so that 
there is no need to switch to a different language.  I would like to make
another point on the fomat of the program structure.  I like the fomat
of F77.  I know exactly where one line begins and ends without having to
search over the page for the semicolon.  Multiple line is not 
a problem in F77 since you can always put a continuation sign at
the 6th column.  

Tang Wong