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

[Ryo ONODERA <ryo_on%yk.rim.or.jp@localhost>]

From: Ryo ONODERA <ryo_on%yk.rim.or.jp@localhost>, Date: Sat, 18 May 2013 
09:08:25 +0900 (JST)

> Hi,
> 
> From: Joerg Sonnenberger <joerg%britannica.bec.de@localhost>, Date: Sat, 18 
> May 2013 01:10:04 +0200
> 
>> On Sat, May 18, 2013 at 07:49:39AM +0900, Ryo ONODERA wrote:
>>> If fortran and fortran77 should use same compiler, what is the mean of
>>> having separate compiler options, fortran and fortran77.
>> 
>> Until recently, requiring a separate compiler would have been way too
>> heavy for something like devel/libtool-base, which is needed by almost
>> anything else. Now that Fortran is split off, the situation is much less
>> dramatic.
> 
> Do you want to say f2c is light package, and g95 is heavy package?
> 
>> If someone really only wants to use Fortran77, f2c is good
>> enough, but given e.g. the inability to build R with it, I find that a
>> bad default choice.
> 
> I have thought when USE_LANGUAGES=fortran, g95 should be used,
> and when USE_LANGUAGES=fortran77, f2c should be used.
> 
> But according to joerg@, fortran77 and fortran cannot coexist as
> pkgsrc compiler, because ABI incompatibility. My understanding is right?
> 
> If fortran77 and fortran cannot coexist as pkgsrc compiler,
> when f2c is default fortran compiler, USE_LANGUAGES=fortran should not
> invoke g95.

Of course, when g95 is default compiler, USE_LANGUAGES=fortran77
should not invoke f2c too.

> By the way,
> I have thought fc2 is only supports fortran77,
> but in pkgsrc/mk/compiler/f2c.mk,
> LANGUAGES.f2c=fortran fortran77 is defined.
> Is this correct?
> 
> --
> Ryo ONODERA // ryo_on%yk.rim.or.jp@localhost
> PGP fingerprint = 82A2 DC91 76E0 A10A 8ABB  FD1B F404 27FA C7D1 15F3