Re: Fortran 9x: time for a decision

[Steven G. Johnson]

Paul Eggert wrote:
> I agree with others that FXX is a bad name for "the latest Fortran
> compiler".  But I'm puzzled: instead of inventing a new name, why not
> use GNU make's names?  Namely, "FC" for the Fortran compiler, and
> "FFLAGS" for the flags to give it.  The Autoconf macro to set FC can
> be called AC_PROG_FC.

I don't have any particular attachment to FXX.  However, I agree with 
John Eaton that FFLAGS is best left as the flags for F77 for backwards 
compatibility reasons.

I have no problem with FCFLAGS etcetera, it just seemed to break the 
pattern with  CCFLAGS/CFLAGS.  If it's decreed GNU-lly correct, so be 
it.  =)

Eric Siegerman wrote:
> That "YY < some_magic_value interpreted as 20YY" was an
> acceptable kludge for Y2K-proofing legacy systems.  But what
> reason is there for incorporating two-digit dates into new
> designs, now that we've seen the grief that can cause?

Because Fortran 77, Fortran 90, and Fortran 95 (not to mention Fortran 
66) *are* pre-Y2K legacy names.  This is the standard terminology 
everyone uses, so I think it makes sense to support it.

Perhaps you'll be happier if it's not presented as an arbitrary year 
(which is not needed), but rather as a canonical version string: 
AC_PROG_FC([VERS], [SEARCH-LIST]), where VERS is simply one of some set 
of known version strings (Fortran [19]77, Fortran [19]90, etcetera, and 
perhaps [19]77, [19]90, [19]95, 2000, etc. for short).

Steven