From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from lists.gentoo.org (pigeon.gentoo.org [208.92.234.80]) by finch.gentoo.org (Postfix) with ESMTP id C77CC13877A for ; Mon, 16 Jun 2014 21:27:02 +0000 (UTC) Received: from pigeon.gentoo.org (localhost [127.0.0.1]) by pigeon.gentoo.org (Postfix) with SMTP id EE9EBE0F57; Mon, 16 Jun 2014 21:26:59 +0000 (UTC) Received: from cantvb.canterbury.ac.nz (cantvb.canterbury.ac.nz [132.181.2.28]) (using TLSv1 with cipher DHE-RSA-AES256-SHA (256/256 bits)) (No client certificate requested) by pigeon.gentoo.org (Postfix) with ESMTPS id 2F857E0F56 for ; Mon, 16 Jun 2014 21:26:59 +0000 (UTC) Received: from CONVERSION-DAEMON.it.canterbury.ac.nz by it.canterbury.ac.nz (PMDF V6.6-x1 #32072) id <01P945NV2EWWC4P9KP@it.canterbury.ac.nz> for gentoo-science@lists.gentoo.org; Tue, 17 Jun 2014 09:26:54 +1300 (NEW ZEALAND DAYLIGHT TIME) Received: from qcd-nzi3.localnet (QCD-nzi3.canterbury.ac.nz [132.181.64.206]) by it.canterbury.ac.nz (PMDF V6.6-x1 #32072) with ESMTPS id <01P945NUV9H2C7K2P7@it.canterbury.ac.nz> for gentoo-science@lists.gentoo.org; Tue, 17 Jun 2014 09:26:54 +1300 (NEW ZEALAND DAYLIGHT TIME) Date: Tue, 17 Jun 2014 09:26:53 +1200 From: =?ISO-8859-1?Q?Fran=E7ois?= Bissey Subject: Re: [gentoo-science] Trouble building scipy-0.14.0 against acml In-reply-to: <539F159A.7010705@arcor.de> To: gentoo-science@lists.gentoo.org Message-id: <2358114.aUMdxTSU4C@qcd-nzi3> Organization: BlueFern Precedence: bulk List-Post: List-Help: List-Unsubscribe: List-Subscribe: List-Id: Gentoo Linux mail X-BeenThere: gentoo-science@lists.gentoo.org Reply-to: gentoo-science@lists.gentoo.org MIME-version: 1.0 Content-type: text/plain; charset=iso-8859-1 Content-transfer-encoding: quoted-printable User-Agent: KMail/4.12.5 (Linux/3.8.13-gentoo; KDE/4.12.5; x86_64; ; ) References: <53959961.1060000@bastian-loeffler.de> <2226924.ppVAimK6bd@qcd-nzi3> <539F159A.7010705@arcor.de> X-Archives-Salt: 3e58729e-628e-4d75-ba3f-a6835a6c956d X-Archives-Hash: 46af1665df2fd4bc79b7f1f7aa08257c OK we have a culprit. You need to eselect a cblas. There is no way that the message you get from pkg-config --libs-only-L cblas blas will be parsed to anything useful. Francois On Mon, 16 Jun 2014 18:04:42 Bastian L=F6ffler wrote: > Hi, >=20 > this is for lapack: >=20 > pkg-config --libs-only-L lapack > -L//opt/acml5.3.1/gfortran64/lib >=20 > cat /usr/lib64/pkgconfig/lapack.pc > prefix=3D//opt/acml5.3.1/gfortran64 > libdir=3D${prefix}/lib > includedir=3D${prefix}/include > Name: acml-gfortran64 > Description: AMD Core Math Library for amd64 CPUs > Version: 5.3.1 > URL: > http://developer.amd.com/tools/cpu-development/amd-core-math-library-= acml/ > Libs: -L${libdir} -lacml > Libs.private: -lm -lrt > Cflags: -I${includedir} >=20 > ls -la /usr/lib64/pkgconfig/lapack.pc > lrwxrwxrwx 1 root root 77 14. Jan 00:04 /usr/lib64/pkgconfig/lapack.p= c > -> > ../../../etc/env.d/alternatives/lapack/_current/usr/lib64/pkgconfig/l= apack.p > c >=20 > and this is >=20 > pkg-config --libs-only-L cblas blas >=20 > Package cblas was not found in the pkg-config search path. > Perhaps you should add the directory containing `cblas.pc' > to the PKG_CONFIG_PATH environment variable > No package 'cblas' found >=20 > I have no cblas eselected. >=20 > Am 16.06.2014 03:27, schrieb Fran=E7ois Bissey: > > And > > pkg-config --libs-only-L cblas blas > > since that's what is actually parsed by the ebuild. Have you > > any cblas eselected? If yes which one. > >=20 > > Francois > >=20 > > On Mon, 16 Jun 2014 13:19:57 Fran=E7ois Bissey wrote: > >> Out of curiosity and to jog my brain can you send the equivalent f= or > >> lapack? > >>=20 > >> Something is definitely going wrong when the site.cfg is generated= . > >>=20 > >> Francois > >>=20 > >> On Sun, 15 Jun 2014 23:40:13 Bastian L=F6ffler wrote: > >>> pkg-config --libs-only-L blas > >>> -L//opt/acml5.3.1/gfortran64/lib > >>>=20 > >>> cat /usr/lib64/pkgconfig/blas.pc > >>> prefix=3D//opt/acml5.3.1/gfortran64 > >>> libdir=3D${prefix}/lib > >>> includedir=3D${prefix}/include > >>> Name: acml-gfortran64 > >>> Description: AMD Core Math Library for amd64 CPUs > >>> Version: 5.3.1 > >>> URL: > >>> http://developer.amd.com/tools/cpu-development/amd-core-math-libr= ary-acm > >>> l/ > >>> Libs: -L${libdir} -lacml > >>> Libs.private: -lm -lrt > >>> Cflags: -I${includedir} > >>>=20 > >>> ls -la /usr/lib64/pkgconfig/blas.pc > >>> lrwxrwxrwx 1 root root 73 14. Jan 00:04 /usr/lib64/pkgconfig/blas= .pc -> > >>> ../../../etc/env.d/alternatives/blas/_current/usr/lib64/pkgconfig= /blas.p > >>> c > >>>=20 > >>> Am 15.06.2014 23:32, schrieb Fran=E7ois Bissey: > >>>> OK so it looks sane, let's do some more sanity check as in the e= mail > >>>> I linked: > >>>> pkg-config --libs-only-L blas > >>>> cat /usr/lib64/pkgconfig/blas.pc > >>>> ls -la /usr/lib64/pkgconfig/blas.pc > >>>>=20 > >>>> On Sun, 15 Jun 2014 23:30:22 Bastian L=F6ffler wrote: > >>>>> Well of course i checked that: > >>>>>=20 > >>>>> eselect blas list > >>>>>=20 > >>>>> Available providers for blas: > >>>>> [1] acml-gfortran64 * > >>>>> [2] acml-gfortran64-openmp > >>>>> [3] eigen > >>>>> [4] reference > >>>>>=20 > >>>>> SO my blas is acml-gfortran64.