On Thu, 11 Dec 2003, Norman Gray wrote:
> Tim,
>
> On Thursday, December 11, 2003, at 10:15 PM, Tim Jenness wrote:
>
> >> By the way, the .F approach for preprocessable Fortran won't work
> >> reliably on HFS+, since file.F and file.f are the same file.... I'd
> >> suggest .fpp as an alternative extension, and just risk folk confusing
> >> it with Fortran++.
> >>
> >
> > Will that be a problem? Since there will not be both a file.F and
> > file.f...
>
> But we'd have to cope with the case where the Fortran compiler couldn't
> cope with a .F file (g77 can, I imagine, but can Sun/DEC/MIPS
> Fortran?), so I'd have to be able to generate the .f file from the .F
> file as a separate step prior to feeding it in to the Fortran compiler.
>
I know that Alpha, Sun and Linux are fine with .F files. Otherwise SURF
would not build. I can't say anything about MIPS compilers though. I did
SURF with .F files before I realised that it wasn't the "approved" way.
[and besides, a .f target from a .fpp file should simply involve the
preprocessor]
> What HFS+ tells us is that, in this day and age, as a general
> principle, it's asking for trouble to rely on case as a distinguishing
> feature of filenames. Even if it _is_ OK in this particular situation,
> relying on case will stuff you up somewhere, somehow.
>
g77 does support .fpp so that's fine with me.
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|