Date: Mon, 4 Jun 2007 09:43:03 +0100 From: Bob Bishop <rb@gid.co.uk> To: Steve Kargl <sgk@troutmask.apl.washington.edu> Cc: ports@freebsd.org, Don Lewis <truckman@freebsd.org>, current@freebsd.org Subject: Re: /usr/share/mk/sys.mk and FORTRAN on -CURRENT Message-ID: <80BDCC4A-97A0-4582-A237-D87B1143130C@gid.co.uk> In-Reply-To: <20070604010942.GA13434@troutmask.apl.washington.edu> References: <200706032347.l53NlXU6055249@gw.catspoiler.org> <20070604010942.GA13434@troutmask.apl.washington.edu>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi,
On 4 Jun 2007, at 02:09, Steve Kargl wrote:
> On Sun, Jun 03, 2007 at 04:47:33PM -0700, Don Lewis wrote:
>> I was taking a look at why math/lapack fails to build on -CURRENT and
>> discovered that the port uses ${FC} to find the FORTRAN compiler. I
>> then discovered that /usr/share/mk/sys.mk defines ${FC} as f77 (or
>> fort77). I don't this is appropriate since f77 has been removed from
>> -CURRENT.
>>
>> I don't know what the proper fix is, but I though I should point
>> out the
>> problem ...
>>
>
> There are two bugs here. 1) FC should be removed.
Why? One might well want the option to use any of the several FORTRAN
compilers from ports.
> 2) The various ports that use Fortran should use gfortran42
> installed with lang/gcc42.
The various ports that use Fortran should be checked to see what
version of FORTRAN they want. I suspect the easiest fix is to have
$FC empty by default and for dependent ports to output a helpful
diagnostic.
A better fix would be to make ports dependent on an appropriate
compiler port, but that's a lot of work and some users will prefer
ifc anyway.
> --
> Steve
--
Bob Bishop +44 (0)118 940 1243
rb@gid.co.uk fax +44 (0)118 940 1295
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?80BDCC4A-97A0-4582-A237-D87B1143130C>