FORTRAN vs. Fortran (was: November 5th is Clang-Day)
sgk at troutmask.apl.washington.edu
Fri Nov 2 13:07:05 UTC 2012
On Fri, Nov 02, 2012 at 08:25:58AM +0000, David Chisnall wrote:
> On 2 Nov 2012, at 08:18, Mehmet Erol Sanliturk wrote:
> > Very many years ago , when 2010 was a very distant future , I do not
> > remember the name of the writer , who wrote approximately :
> > "In 2010 , there will be Fortran , but a Fortran which may be different ."
> I remember a talk in the mid '90s by someone from Sun's HPC team
> where he said 'I don't know what the syntax or semantics of the
> language we will be using for HPC in 20 years time will be, but
> I do know one thing about it: it will be called Fortran'
> Although the response to GCC's recent decision to drop support
> for Fortran 77 showed that that language will probably be called Fortran 77...
GCC did not drop Fortran 77. When GCC moved from the 3.x series
to the 4.x series, it introduced the use of gimple and tree-ssa.
No one ported g77 to use gimple and tree-ssa, so g77 was replaced
by a completely new frontend, which is called gfortran and started
life as a Fortran 95 compiler. One admirable objective of J3, the
Fortran standardization committee, is that it strives for backwards
compatibility to previous standards. So, if you have a valid
Fortran 77 code, it will in all likelihood be a validate Fortran
2008 program. Fortran 95 deleted 10 features from the language;
however, every compiler that I've used still supports those features.
In regards to HPC and Fortran, Fortran 2008 introduced this wonderful
feature called co-arrays. One can read about gfortran's progress with
its implementation of co-arrays at http://gcc.gnu.org/wiki/Coarray
More information about the freebsd-current