tech-toolchain archive

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index][Old Index]

Re: put fortran back in base?



David Holland <dholland-tech%netbsd.org@localhost> writes:

> It was brought to my attention that there was an old PR from 2009
> suggesting adding fortran back to base (by including the gcc fortran
> frontend) because otherwise various messy things happened with pkgsrc
> packages using fortran.
>
> Those particular pkgsrc issues no longer apply, but others do, and
> there's been some talk about that lately -- the default in pkgsrc is
> g95, but g95 is old, dead upstream, and not without issues; the
> alternative is building a pkgsrc gcc with fortran but this, in
> addition to being a general pain, potentially leads to problems with
> the pkgsrc gcc and base gcc not being quite compatible.
>
> Meanwhile, much as we'd like to pretend fortran is dead, it's not and
> that's not going to change in the foreseeable future. It's certainly
> reasonable to include a fortran compiler in base, as well as
> historically accepted. I don't know if we ought to or not, but I think
> it's worth noting the pros and cons.

I agree that fortran is very much not dead.  My guess is that enabling
gfortran in our in-tree gcc builds will be only a fairly small bit of
size and build time, and it's in the comp set that people on embedded
systems will exclude anyway.  Arguably it could be a MK variable,
defaulting to yes, which also makes it easier to avoid.

All in all, I think it would be good to have fortran in base because of
the combination of it being
  reliable
  historical
  useful
  coming as part of a package we already have in base

I also wonder if clang does fortran in a reasonable and satisfying way;
a quick web search turns up 'dragonegg' which is a hybrid gcc/clang
system that is not quite maintained and has some issues.

Attachment: signature.asc
Description: PGP signature



Home | Main Index | Thread Index | Old Index