diff options
| author | Edwin Groothuis <edwin@FreeBSD.org> | 2006-01-03 04:37:24 +0000 |
|---|---|---|
| committer | Edwin Groothuis <edwin@FreeBSD.org> | 2006-01-03 04:37:24 +0000 |
| commit | f0ac2b0eed66cd5fc16e38a187c74f28eec296e3 (patch) | |
| tree | 22c9e4d49dfe523027fbcc7feaa552973b7f9de9 /math | |
| parent | efe276b7140eb56300c6a098100619ffd3f1839e (diff) | |
| download | ports-f0ac2b0eed66cd5fc16e38a187c74f28eec296e3.tar.gz ports-f0ac2b0eed66cd5fc16e38a187c74f28eec296e3.zip | |
Make math/arpack depend less on math/atlas
Atlas is great in the platforms where it's available, however
it is not being packaged (for good reasons) anymore. arpack
originally used blas for testing and since we are not really
linking any version of blas in the package it doesn't make
any difference what we use. It seems better idea to use
blas for testing since it's faster to build and because
this also means we can package arpack without building
ATLAS, The resulting package will work with BLAs or ATLAS
just the same and in any case the BLAS_LIBS variable can
now be overridden to test with atlas if it's so desired.
This change doesn't change the package so I think we should
leave the PORTVERSION untouched.
PR: ports/91251
Submitted by: Pedro F. Giffuni <giffunip@asme.org>
Notes
Notes:
svn path=/head/; revision=152643
Diffstat (limited to 'math')
| -rw-r--r-- | math/arpack/Makefile | 4 |
1 files changed, 2 insertions, 2 deletions
diff --git a/math/arpack/Makefile b/math/arpack/Makefile index 1524c8d0a2f4..790a2bde5238 100644 --- a/math/arpack/Makefile +++ b/math/arpack/Makefile @@ -22,12 +22,12 @@ MAINTAINER= ports@FreeBSD.org COMMENT= Argand Library: large eigenvalue subroutines (serial version) BUILD_DEPENDS= ${LOCALBASE}/bin/f2c:${PORTSDIR}/lang/f2c \ - ${LOCALBASE}/lib/libf77blas.a:${PORTSDIR}/math/atlas + ${LOCALBASE}/lib/libblas.a:${PORTSDIR}/math/blas WRKSRC= ${WRKDIR}/ARPACK ALL_TARGET= lib -BLAS_LIBS= ${LOCALBASE}/lib/libf77blas.a ${LOCALBASE}/lib/libatlas.a +BLAS_LIBS?= ${LOCALBASE}/lib/libblas.a INSTALLS_SHLIB= yes USE_REINPLACE= yes |