On Thu, 5 Feb 2004 08:05:21 -0800, Richard Maine <[log in to unmask]>
wrote:
> 3. Style changes.
There's a lot in this simple statement. Writing (and maintaining) "clean"
Fortran 95 style programs is much simpler than "mixed-language" (77 and 95)
programming. In the later case you need to know not only both languages,
but
the subtle mechanisms that make them work together (for example, what an
explicit interface is and when you need them; you don't need to know that
if you just start with everything in modules, or you don't use any 95
stuff).
In that context I can understand a want for the ideal automatic tool that
can
make everything "95". The tool should convert 77 code so that say, only an
F
subset knowledge is required on the maintainer. That would be a meaningful
conversion I think. Whether that is possible depends a lot on how cleanly
the input code was written I guess.
Or providing 95 wrappers to 77 routines to cleanly separate the two is also
a good option. That way only "95", "77", and "interfacing" knowledge is
necessary.
The alternate "mixed-language" approach (where you need to know everything
else
in addition to that) is an option for the experts, not for the guys fresh
out of
a one-week Fortran 95 course.
Yasuki Arasaki
|