Re: ping: updating Fortran support for f9x?

[Steven G. Johnson]

Hi Philippe, thanks for your comments.

On Tue, 8 Jul 2003, Philippe Trottier wrote:
>       The detection of GNU fortran is done by passing this code to the 
> compiler with the extention .F

This is orthogonal to F9x support; the current autoconf uses the same
behavior.

(This test is mostly there for legacy reasons anyway; I'm not sur

>       Then Absoft f77 and f95 will require f95 -V valid_source.f95 to
> return the version number ... so is f77, so the version check is also
> not working. maybe add a case with -V and -v with

What version check?  There is only a check to get verbose output (for
AC_F77_LIBRARY_LDFLAGS), unless I'm misremembering; is this what you are
referring to?  Does AC_F77_LIBRARY_LDFLAGS work with the Absoft compiler?

Again, this is orthogonal to F9x support, since the verbose-output macro
is directly copied from the current autoconf.

>       Some compiler behaves a bit differently when called with
> different options, should the previous test use the AC_FC_SRCEXT ?
> making this before the compiler check ?

AC_FC_SRCEXT is for the user to specify a desired source extension; if
it's not specified, .f is assumed on the KISS principle (there doesn't
seem to be any better choice).

It's certainly possible for the user to call AC_FC_SRCEXT before calling
AC_FC_LIBRARY_LDFLAGS, if that's what you're worried about.

Steven