Path: utzoo!attcan!uunet!lll-winken!lll-tis!helios.ee.lbl.gov!pasteur!ames!mailrus!uflorida!haven!uvaarpa!hudson!bessel.acc.Virginia.EDU!gl8f
From: gl8f@bessel.acc.Virginia.EDU (Greg Lindahl)
Newsgroups: comp.lang.fortran
Subject: Re: Fortran vs C for computations
Message-ID: <559@hudson.acc.virginia.edu>
Date: 15 Sep 88 23:47:09 GMT
References: <391@quintus.UUCP> <3494@lanl.gov> <1510@ficc.uu.net>
Sender: news@hudson.acc.virginia.edu
Reply-To: gl8f@bessel.acc.Virginia.EDU (Greg Lindahl)
Organization: Dept. of Astronomy, University of Virginia
Lines: 18

In article <1510@ficc.uu.net> peter@ficc.uu.net (Peter da Silva) writes:
>
> [stuff deleted]
>
> There is no standard program available with all Fortran compilers that
> does preprocessing of the program (conditional compilation, etc).

One large image processing package (AIPS) is written in FORTRAN and is
portable across all kinds of environments -- VAX/VMS, Unix V and 4.X,
Crays (COS and Unicos), etc. It was done without preprocessing at all, 
except for certain operating systems (UNIX) which have particurally
brain-damaged FORTRAN compilers (no INCLUDE statement). (I'm not claiming
that INCLUDE is standard, merely that is was useful to the AIPS programmers.)
They achieved portability by isolating operating system dependent code
into routines which were rewritten for each operating system. This isn't
the most elegant approach, but it works without preprocessing.

Greg Lindahl                              internet:  gl8f@virginia.edu
U Va Dept. of Astronomy                   bitnet:    gl8f@virginia.bitnet