Bugzilla Bug 97417

Working ebuilds of sys-cluster/lam-mpi do not build with fortran (f77) support.
 This includes lam-7.0.6 and lam-7.1.  lam-7.1.1 does attempt to build fortran
support, but the entire ebuild doesn't work, and it doesn't follow the fortran
use variable.

I have a patched version of the ebuild (which I will attach) for version 7.0.6
which checks for the fortran use variable similar to the way version 7.0.4-r1
worked and the builds fortran support with g77.

Reproducible: Always
Steps to Reproduce:
1. emerge lam-mpi
2. mpif77 -showme

Actual Results:  
mpif77 fails

Expected Results:  
mpif77 should not fail.

Created an attachment (id=62260) [edit]
a patch to the ebuild for lam-mpi-7.0.6 to enable fortran use variable

This is the patch I mentioned in the bug report.

(From update of attachment 62260 [edit])
>--- /usr/portage/sys-cluster/lam-mpi/lam-mpi-7.0.6.ebuild	2005-02-07 07:12:06.000000000 -0500
>+++ /usr/local/portage/sys-cluster/lam-mpi/lam-mpi-7.0.6.ebuild	2005-06-29 15:27:06.000000000 -0400
>@@ -2,7 +2,7 @@
> # Distributed under the terms of the GNU General Public License v2
> # $Header: /var/cvsroot/gentoo-x86/sys-cluster/lam-mpi/lam-mpi-7.0.6.ebuild,v 1.6 2005/02/07 12:01:47 eradicator Exp $
> 
>-IUSE="crypt"
>+IUSE="crypt fortran"
> 
> MY_P=${P/-mpi}
> S=${WORKDIR}/${MY_P}
>@@ -39,6 +39,12 @@ src_compile() {
> 		myconf="--with-rsh=rsh"
> 	fi
> 
>+	if use fortran; then
>+		myconf="${myconf} --with-fc=g77"
>+	else
>+		myconf="${myconf} --without-fc"
>+	fi
>+
> 	econf \
> 		--sysconfdir=/etc/lam-mpi \
> 		--enable-shared \
>@@ -48,7 +54,7 @@ src_compile() {
> 		--disable-checking  \
> 		--enable-cstdio=stdio \
> 		--with-system-zlib \
>-		--without-fc \
> 		${myconf} || die
> 
> 	# sometimes emake doesn't finish since it gets ahead of itself :)

Sorry about that.  There was a mistake in my patch, and I didn't know how to
edit it without turning it into a comment instead of an attachment.

fixed in cvs.

This commit is not correct. It assumes that the Fortran compiler is g77 when
that is certainly not always the case, particularly with the advent of gcc4.

Robin, you may wish to take more advantage of the fortran eclass, since lam-mpi
already inherits it.

ok fixed now.

Thanks!