[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [autoconf] Fortran 90

From: Lassi A. Tuura
Subject: Re: [autoconf] Fortran 90
Date: Wed, 08 May 2002 11:22:41 +0200

> Are there people who are going to use more than one dialect at a time
> rather than using only the best compiler available?

Some projects still require strict adherence to f77: no post-f77
features are wanted.  OTOH, many F77 compilers grok F95 as well -- if
you feed them source with F90 or F95 constructs in it, they might well
compile it.  I am not sure if that applies to modules and such, but it
often works for things like A = B + C where A, B and C are matrices. 
We've had quite a lot of code like that pass unnoticed until someone
compiled it on a strict f77 compiler.  And then there's of course the
Windows FORTRAN compilers (MS? Compaq?) that won't do preprocessing
and/or are from a different planet when it comes to command line
interface -- at least they used to (IIRC they wouldn't accept forward
slashes anywhere on the command line except for options, e.g. a forward
slash in -L or -I options or even the source file name would cause most
severe grief -- normally not a problem with Microsoft compilers).

Certainly in a project I was involved in most of the code -- a few
million lines -- everything was required to be F77.  Later an allowance
was made for F90 in restricted parts of the system, so yes, there would
have been need for different language levels (the build system did
involve autoconf-based configure scripts but is no longer used).

In my present project I am much less familiar with the FORTRAN side but
I assume the code is 100% F77.  But I don't think it uses autoconf at
all so that's just anecdotal evidence.

I remember past discussions on this topic over here.  It would be wise
to check the mail archives for all the oddities.

Waste no more time arguing what a good man should be.  Be one.
        --Marcus Aurelius

reply via email to

[Prev in Thread] Current Thread [Next in Thread]