Re: math/blas and math/lapack broken: cannot find -lf95

[Joerg Sonnenberger <joerg%britannica.bec.de@localhost>]

On Wed, May 15, 2013 at 11:22:50PM +0200, Thomas Klausner wrote:
> On Wed, May 15, 2013 at 10:09:38PM +0200, Joerg Sonnenberger wrote:
> > On Wed, May 15, 2013 at 05:21:37PM +0200, Thomas Klausner wrote:
> > > On Sun, May 12, 2013 at 03:12:44PM +0900, Ryo ONODERA wrote:
> > > > Old libtool-base has
> > > > 
> > > > USE_LANGUAGES=   c c++ fortran77
> > > > 
> > > > in its Makefile.
> > > > And libtool-fortran has
> > > > 
> > > > USE_LANGUAGES=  c c++ fortran
> > > > 
> > > > in its Makefile.
> > > > 
> > > > Should the following patch be committed?
> > > 
> > > Thank you, I've committed a similar patch.
> > 
> > Thanks for honouring OWNER. Please revert, the dependency on fortran was
> > intentional.
> 
> Well, you broke it and didn't reply to the breakage, nor did you
> comment negatively on the patch.

I didn't comment on it at all. Which should point out that I either
didn't have time to deal with it or didn't read it at all.

Now about the content of the reverted change: it is wrong. It is
artifically pessimizing the build for ancient Fortran, which is clearly
not useful for most of the things that depend on Fortran. So the real
question is now, whether there is any point in keeping Fortran77 support
as individual option, given that you can't easily mix different Fortran
implementations due to the general lack of a stable ABI. Use a
consistent Fortran compiler for "fortran" and "fortran77" and this
problem doesn't appear.

Joerg