A second PR (standards/19640).  Again, I am sending a copy of it to the
right place, the tech-userlevel mailing list, for discussion.  I really
appreciate any comments on this proposal.


Proposal:  Moving g77 to the ports

The GNU FORTRAN compiler has stayed in release 0.5.x since mid-90.
I propose moving it to the ports system.  I will try to outline
the reasons for this proposal.

1. g77 stays in release 0.5.x since mid-90.

   That is not a bad thing!  All of us like software that is stable,
   we are running BSDs for this reason, amongst other things.  But
   when a package has serious problems and it is not fixed, stability
   is bad.

2. Even GNU is dropping it.

   GNU is developing a new FORTRAN compiler.  g77 is not the best
   compiler in the GNU Compiler Collection (GCC) suite.

3. It is something between a FORTRAN 77 and a FORTRAN 90 compiler.

   It does not follow standards, it is a FORTRAN 77 compiler
   that supports "some FORTRAN 90 extensions".  I think that it
   is not as good as other GNU developments, and it does not
   follow standards in the right way.  I think that it has not
   the quality level required to be a part of a BSD system.

   On the other hand, gcc and g++ are high-quality products.

4. It is not being widely used.

   It is not a required component of a BSD system and it is not
   used when build a BSD system from the source code.  Moving
   it to the ports collection is not bad.  Those changes from
   the base system to the ports collection happened with other
   packages (like most of the a.out stuff) on some BSD systems.

   Most people that wants a FORTRAN compiler for serious work
   runs other products.  For example, the Condensed Matter group
   at the Physics Department of our University runs a commercial
   product in the Linux cluster.  The dropped g77 in 1996.

Same about new compilers provided in the GCC suite, that are being
provided in the latest releases of the Compiler Collection.  Those
compilers (e.g. java) can be provided too as ports.

Thanks again!
Igor.

-- 
Igor Sobrado, UK34436 - sobrado@acm.org