I should also add that I'm willing to implement this in autoconf (say
next weekend; I'll be gone this week) if Akim et al. agree that it (or
something similar) is the way to go.
Regarding automake, I would think that the ordinary behavior would be to
use $F77 for {.f, .F, .for, .f77} and $FXX for everything else (.f90,
.f95, etc. [*]). However, if the configure.ac calls AC_PROG_FXX and not
AC_PROG_F77, then automake should use $FXX for everything. Rationale:
-- if the user calls AC_PROG_F77, $F77 should be used for backwards
compatibility, and to provide a way to use the F77 compiler for old
codes in mixed-dielect projects.
-- it is good to support using $FXX for everything, especially for
cases where the F77 and FXX compilers may not be compatible.
-- it is good to allow users of the simple .f extension to employ the
newer compilers and language features if they choose. Free .f from the
tyranny of 1977! =)
-- AC_PROG_FXX(77) will give the F77 compiler anyway.
Steven
[*] Does anyone know what filename extension is proposed, if any, for
Fortran 2000? .f2k? I don't know how people can continue this trend
indefinitely, at least with the Windows cultural bias of at most
3-character extensions.
- Fortran 9x: time for a decision Steven G. Johnson
- Re: Fortran 9x: time for a decision Steven G. Johnson
- Re: Fortran 9x: time for a decision Paul Eggert
- Re: Fortran 9x: time for a decision Steven G. Johnson
- Re: Fortran 9x: time for a decision Paul Eggert
- Re: Fortran 9x: time for a decisio... Steven G. Johnson
- Re: Fortran 9x: time for a dec... Earnie Boyd
- Re: Fortran 9x: time for a... Kate Hedstrom
- Re: Fortran 9x: time for a... Steven G. Johnson
- Re: Fortran 9x: time for a decision Steven G. Johnson
- Re: Fortran 9x: time for a decision Eric Siegerman
- Re: Fortran 9x: time for a decision Kate Hedstrom