Date: Sat, 27 May 2006 06:01:52 +0200 (CEST) From: <pfgshield-freebsd@yahoo.com> To: freebsd-current@freebsd.org, Steve Kargl <sgk@troutmask.apl.washington.edu> Subject: Re: [RFC] Removal of Fortran from the base system Message-ID: <20060527040152.89907.qmail@web32714.mail.mud.yahoo.com>
next in thread | raw e-mail | index | archive | help
Hi;
FWIW, I think I am right now the only user of gfortran in the ports tree as I
use it for the elmer* ports and really soon for MUMPS (which is finished but
being tested).
First of all I should mention g77 and gfortran are not ABI compatible by
default, and specifying -ff2c to gfortran is not always easy; this has meant,
for example, that it was necessary to add a suboptimal package called
elmer-mathlibs (with blas, lapack and arpack/parpack) just for elmer.
Libraries generated with g77 will not interoperate with the gfortran stuff and
some packages (math/blacs comes to mind) will need to be patched to work with
gfortran. That said, gfortran41 is a pretty good replacement to g77.
I wouldn't count much with Intel's compiler since it's a binary that we can't
redistribute and has platform limitations. g95 is better and much more used but
it is not easy to package (I tried on amd64).
I don't have clear atm if it's a good idea to remove fortran or not but NOT
having a fortran compiler in the base system would be extremely weird. A couple
of questions to think about:
- Will we need a package with the shared libraries that come with gfortran?
many packages depend on blas/atlas and other libraries that need fortran but a
run dependency on the compiler package would be excessive IMHO.
- Perhaps we could remove of the C compiler too? not everyone builds
kernels/ports and the gfortran compiler package happens to include a good C
compiler ;-).
cheers,
Pedro.
Chiacchiera con i tuoi amici in tempo reale!
http://it.yahoo.com/mail_it/foot/*http://it.messenger.yahoo.com
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20060527040152.89907.qmail>