Re: /ports/math/blas does not build
On Tue, Aug 17, 2010 at 11:39:41AM -0400, as...@spamcop.net wrote: > Hi, > > So I have been struggling with blas in a -current i386. Vaguely, the > fortran files and the man files do not get "installed" so things do > not get built. Now I broke down and did a pkg_add and it was not so > much maybe that blas got installed, but that a lot of file got put > on my machine and a lot of fortran compiling went on the next time I > tried a ports build. Having played with downloading stuff from > libnet, I sort of think the distfiles are really not what we are > expecting. This is not anything I am able to act on, but I thought > I would mention it. If I was going to try to reproduce, I think I > would erase everything in /usr/ports/pobj. Looks like you're very confused. There is absolutely nothing we don't know in there. Like, hey fortran got removed from base for various reasons, it was discussed to death.
Re: /ports/math/blas does not build
On 08/17/10 17:39, as...@spamcop.net wrote: Hi, So I have been struggling with blas in a -current i386. Vaguely, the fortran files and the man files do not get "installed" so things do not get built. Now I broke down and did a pkg_add and it was not so much maybe that blas got installed, but that a lot of file got put on my machine and a lot of fortran compiling went on the next time I tried a ports build. Having played with downloading stuff from libnet, I sort of think the distfiles are really not what we are expecting. This is not anything I am able to act on, but I thought I would mention it. If I was going to try to reproduce, I think I would erase everything in /usr/ports/pobj. For what it is worth, it looks like there is even newer code on libnet than we are using. did you apply my patch ? -- Thomas de Grivel http://b.lowh.net/billitch
Re: /ports/math/blas does not build
On 2010/08/17 11:39, as...@spamcop.net wrote: > Hi, > > So I have been struggling with blas in a -current i386. Vaguely, the > fortran files and the man files do not get "installed" so things do > not get built. Now I broke down and did a pkg_add and it was not so > much maybe that blas got installed, but that a lot of file got put on > my machine and a lot of fortran compiling went on the next time I > tried a ports build. Having played with downloading stuff from > libnet, I sort of think the distfiles are really not what we are > expecting. This is not anything I am able to act on, but I thought I > would mention it. If I was going to try to reproduce, I think I > would erase everything in /usr/ports/pobj. I don't really understand what you're saying here... > For what it is worth, it looks like there is even newer code on > libnet than we are using. Yes, it breaks some things iirc.
/ports/math/blas does not build
Hi, So I have been struggling with blas in a -current i386. Vaguely, the fortran files and the man files do not get "installed" so things do not get built. Now I broke down and did a pkg_add and it was not so much maybe that blas got installed, but that a lot of file got put on my machine and a lot of fortran compiling went on the next time I tried a ports build. Having played with downloading stuff from libnet, I sort of think the distfiles are really not what we are expecting. This is not anything I am able to act on, but I thought I would mention it. If I was going to try to reproduce, I think I would erase everything in /usr/ports/pobj. For what it is worth, it looks like there is even newer code on libnet than we are using.
Re: [patch] math/blas-1.0p4 (was Re: /ports/math/blas does not build)
On Sat, Aug 14, 2010 at 10:15:09AM +0100, Stuart Henderson wrote: > On 2010/08/14 11:06, Steven Mestdagh wrote: > > > > Thanks for noticing that. I'll take a look at this later. > > Looks like they bundled a makefile with the code so we may as well use that. > > You should be able to build the in tree port with distfiles from > > ftp.openbsd.org. > > It seems that ftp.openbsd.org has mirrored the newer version.. > You can fetch the old distfiles from http://junkpile.org/blas-1.0/ > (I thought we had some symlink magic to take care of this situation > but I might be mistaken..) make checksum REFETCH=true does work, so yeah, mirroring is effective ;-)
Re: [patch] math/blas-1.0p4 (was Re: /ports/math/blas does not build)
On 2010/08/14 11:06, Steven Mestdagh wrote: > > Thanks for noticing that. I'll take a look at this later. > Looks like they bundled a makefile with the code so we may as well use that. > You should be able to build the in tree port with distfiles from > ftp.openbsd.org. It seems that ftp.openbsd.org has mirrored the newer version.. You can fetch the old distfiles from http://junkpile.org/blas-1.0/ (I thought we had some symlink magic to take care of this situation but I might be mistaken..)
Re: [patch] math/blas-1.0p4 (was Re: /ports/math/blas does not build)
Thomas de Grivel [2010-08-14, 10:41:37]:
> On 08/14/10 08:14, as...@spamcop.net wrote:
> >I have not figured out how to cut and paste from xterm and I expect that
> >attachments are not the thing to do. Thus no nice build outputs. My
> >-current is say two days old. There are four patches and none of the
> >patches cleanly apply.
> >
> >I note it seems to be almost 20 year old code, fortran, and openoffice3
> >build is dependent on it.
>
> py-gtk is dependent on it I think.
> Do you mean copy / paste like, mouse select / button 2 ?
> Attachments are ok.
>
> Attached is a patch for math/blas reflecting the silent changes from
> (unversioned) upstream tarball.
>
> I do not use blas, i just tried to fix the build so please review !
Thanks for noticing that. I'll take a look at this later.
Looks like they bundled a makefile with the code so we may as well use that.
You should be able to build the in tree port with distfiles from
ftp.openbsd.org.
>
> Cheers,
>
> --
> Thomas de Grivel
> http://www.lowh.net/
> Index: Makefile
> ===
> RCS file: /cvs/ports/math/blas/Makefile,v
> retrieving revision 1.14
> diff -u -p -r1.14 Makefile
> --- Makefile 18 Jul 2010 10:30:41 - 1.14
> +++ Makefile 13 Aug 2010 15:22:49 -
> @@ -5,7 +5,7 @@ COMMENT= Basic Linear Algebra Subprogram
> VERSION= 1.0
> DISTNAME=blas
> PKGNAME= ${DISTNAME}-${VERSION}
> -REVISION=3
> +REVISION=4
> SHARED_LIBS= blas1.0
>
> CATEGORIES= math
> @@ -25,7 +25,7 @@ MASTER_SITES= ftp://ftp.netlib.org/blas/
> MASTER_SITES0= ${MASTER_SITES:S/blas/lapack/}
> EXTRACT_SUFX=.tgz
> DIST_SUBDIR= ${DISTNAME}-${VERSION}
> -WRKDIST= ${WRKDIR}
> +WRKDIST= ${WRKDIR}/BLAS
>
> MODULES= fortran
> MODFORTRAN_WANTG77= Yes
> @@ -37,7 +37,7 @@ MAKE_ENV= SHLIB_MAJOR=${LIBblas_VERSION:
> FC=${FC}
> FAKE_FLAGS= LIBDIR=${LOCALBASE}/lib DEBUGLIBS=no
>
> -MANSRC= ${WRKDIR}/${DISTNAME}/man/manl
> +MANSRC= ${WRKDIR}/lapack-3.2.0/manpages/blas/man/manl
>
> # See bsd.lib.mk:162
> .if ${MACHINE_ARCH:Mmips64*}
> Index: distinfo
> ===
> RCS file: /cvs/ports/math/blas/distinfo,v
> retrieving revision 1.4
> diff -u -p -r1.4 distinfo
> --- distinfo 5 Apr 2007 16:20:05 - 1.4
> +++ distinfo 13 Aug 2010 15:22:49 -
> @@ -1,10 +1,10 @@
> -MD5 (blas-1.0/blas.tgz) = WBHks9/6ehCVar+SPOwb1A==
> -MD5 (blas-1.0/manpages.tgz) = UO+rbNc6lClYT38VN/Fyfw==
> -RMD160 (blas-1.0/blas.tgz) = 0s6jrmFWtvWD+uXlkGlf+zBqu4A=
> -RMD160 (blas-1.0/manpages.tgz) = lwTUbBejKMZpB8K4c9bfJOQtDyU=
> -SHA1 (blas-1.0/blas.tgz) = q3P8xp0MdIiK6memmPYM2RsNOvI=
> -SHA1 (blas-1.0/manpages.tgz) = c5dcLZ3Q8A8kqK719abmZ1J9L9M=
> -SHA256 (blas-1.0/blas.tgz) = vQYqeeNamoNc/Y6uvmJiu7GVYQM2zwdDXNK33yAL8IQ=
> -SHA256 (blas-1.0/manpages.tgz) = AuMMujV621/rSzipdH4eNjNzKqSmjDJsvYIg/UPIU8Y=
> -SIZE (blas-1.0/blas.tgz) = 94981
> -SIZE (blas-1.0/manpages.tgz) = 721308
> +MD5 (blas-1.0/blas.tgz) = fmr3AiRA2GiNFr6G1V+zWA==
> +MD5 (blas-1.0/manpages.tgz) = FFAHyrkVUEyuw4IolGKhZg==
> +RMD160 (blas-1.0/blas.tgz) = 6lx/7V/j2t82Aj8O4SIrUgAHVFk=
> +RMD160 (blas-1.0/manpages.tgz) = nxByf3pEMoevXNm11mAMlyeQWXE=
> +SHA1 (blas-1.0/blas.tgz) = Cuyk7UXy5VGZd3R+1rz4DRsDNdI=
> +SHA1 (blas-1.0/manpages.tgz) = xVgHPgWxaZi14/LGV9D6z1YuxVE=
> +SHA256 (blas-1.0/blas.tgz) = vC8liYFBw+2VE6vjs/FeAPDS6Igcfya3SVDN7kX7VB0=
> +SHA256 (blas-1.0/manpages.tgz) = ysI05ZfyuTiL89PAVJuKVr84Z8bFVNhEu4OQOZc4YfY=
> +SIZE (blas-1.0/blas.tgz) = 98957
> +SIZE (blas-1.0/manpages.tgz) = 1016997
> Index: patches/patch-blas_man_manl_cscal_l
> ===
> RCS file: patches/patch-blas_man_manl_cscal_l
> diff -N patches/patch-blas_man_manl_cscal_l
> --- patches/patch-blas_man_manl_cscal_l 5 Sep 2007 15:32:19 -
> 1.2
> +++ /dev/null 1 Jan 1970 00:00:00 -
> @@ -1,9 +0,0 @@
> -$OpenBSD: patch-blas_man_manl_cscal_l,v 1.2 2007/09/05 15:32:19 steven Exp $
> blas/man/manl/cscal.l.orig Sat Apr 1 19:54:03 2006
> -+++ blas/man/manl/cscal.lSat Apr 1 19:54:11 2006
> -@@ -1,4 +1,4 @@
> --.SH NAME
> -+.TH CSCAL 3f
> -
> - .SH SYNOPSIS
> - .TP 31
> Index: patches/patch-blas_man_manl_dcabs1_l
> ===
> RCS file: patches/patch-blas_man_manl_dcabs1_l
> diff -N patches/patch-blas_man_manl_dcabs1_l
> --- patches/patch-blas_man_manl_dcabs1_l 5 Sep 2007 15:32:19 -
> 1.2
> +++ /dev/null 1 Jan 1970 00:00:00 -
> @@ -1,9 +0,0 @@
> -$OpenBSD: patch-blas_man_manl_dcabs1_l,v 1.2 2007/09/05 15:32:19 steven Exp $
> blas/man/manl/dcabs1.l.orig Sat Apr 1 19:54:21 2006
> -+++ blas/man/manl/dcabs1.l Sat Apr 1 19:54:33 2006
> -@@ -1,4 +1,4 @@
> --.SH NAME
> -+.TH DCABS1 3f
> -
> - .SH SYNOPSIS
> - .TP 17
> Index: patches/patch-blas_man_manl_zrotg_l
[patch] math/blas-1.0p4 (was Re: /ports/math/blas does not build)
On 08/14/10 08:14, as...@spamcop.net wrote:
I have not figured out how to cut and paste from xterm and I expect that
attachments are not the thing to do. Thus no nice build outputs. My
-current is say two days old. There are four patches and none of the
patches cleanly apply.
I note it seems to be almost 20 year old code, fortran, and openoffice3
build is dependent on it.
py-gtk is dependent on it I think.
Do you mean copy / paste like, mouse select / button 2 ?
Attachments are ok.
Attached is a patch for math/blas reflecting the silent changes from
(unversioned) upstream tarball.
I do not use blas, i just tried to fix the build so please review !
Cheers,
--
Thomas de Grivel
http://www.lowh.net/
Index: Makefile
===
RCS file: /cvs/ports/math/blas/Makefile,v
retrieving revision 1.14
diff -u -p -r1.14 Makefile
--- Makefile18 Jul 2010 10:30:41 - 1.14
+++ Makefile13 Aug 2010 15:22:49 -
@@ -5,7 +5,7 @@ COMMENT=Basic Linear Algebra Subprogram
VERSION= 1.0
DISTNAME= blas
PKGNAME= ${DISTNAME}-${VERSION}
-REVISION= 3
+REVISION= 4
SHARED_LIBS= blas1.0
CATEGORIES=math
@@ -25,7 +25,7 @@ MASTER_SITES= ftp://ftp.netlib.org/blas/
MASTER_SITES0= ${MASTER_SITES:S/blas/lapack/}
EXTRACT_SUFX= .tgz
DIST_SUBDIR= ${DISTNAME}-${VERSION}
-WRKDIST= ${WRKDIR}
+WRKDIST= ${WRKDIR}/BLAS
MODULES= fortran
MODFORTRAN_WANTG77=Yes
@@ -37,7 +37,7 @@ MAKE_ENV= SHLIB_MAJOR=${LIBblas_VERSION:
FC=${FC}
FAKE_FLAGS=LIBDIR=${LOCALBASE}/lib DEBUGLIBS=no
-MANSRC=${WRKDIR}/${DISTNAME}/man/manl
+MANSRC=${WRKDIR}/lapack-3.2.0/manpages/blas/man/manl
# See bsd.lib.mk:162
.if ${MACHINE_ARCH:Mmips64*}
Index: distinfo
===
RCS file: /cvs/ports/math/blas/distinfo,v
retrieving revision 1.4
diff -u -p -r1.4 distinfo
--- distinfo5 Apr 2007 16:20:05 - 1.4
+++ distinfo13 Aug 2010 15:22:49 -
@@ -1,10 +1,10 @@
-MD5 (blas-1.0/blas.tgz) = WBHks9/6ehCVar+SPOwb1A==
-MD5 (blas-1.0/manpages.tgz) = UO+rbNc6lClYT38VN/Fyfw==
-RMD160 (blas-1.0/blas.tgz) = 0s6jrmFWtvWD+uXlkGlf+zBqu4A=
-RMD160 (blas-1.0/manpages.tgz) = lwTUbBejKMZpB8K4c9bfJOQtDyU=
-SHA1 (blas-1.0/blas.tgz) = q3P8xp0MdIiK6memmPYM2RsNOvI=
-SHA1 (blas-1.0/manpages.tgz) = c5dcLZ3Q8A8kqK719abmZ1J9L9M=
-SHA256 (blas-1.0/blas.tgz) = vQYqeeNamoNc/Y6uvmJiu7GVYQM2zwdDXNK33yAL8IQ=
-SHA256 (blas-1.0/manpages.tgz) = AuMMujV621/rSzipdH4eNjNzKqSmjDJsvYIg/UPIU8Y=
-SIZE (blas-1.0/blas.tgz) = 94981
-SIZE (blas-1.0/manpages.tgz) = 721308
+MD5 (blas-1.0/blas.tgz) = fmr3AiRA2GiNFr6G1V+zWA==
+MD5 (blas-1.0/manpages.tgz) = FFAHyrkVUEyuw4IolGKhZg==
+RMD160 (blas-1.0/blas.tgz) = 6lx/7V/j2t82Aj8O4SIrUgAHVFk=
+RMD160 (blas-1.0/manpages.tgz) = nxByf3pEMoevXNm11mAMlyeQWXE=
+SHA1 (blas-1.0/blas.tgz) = Cuyk7UXy5VGZd3R+1rz4DRsDNdI=
+SHA1 (blas-1.0/manpages.tgz) = xVgHPgWxaZi14/LGV9D6z1YuxVE=
+SHA256 (blas-1.0/blas.tgz) = vC8liYFBw+2VE6vjs/FeAPDS6Igcfya3SVDN7kX7VB0=
+SHA256 (blas-1.0/manpages.tgz) = ysI05ZfyuTiL89PAVJuKVr84Z8bFVNhEu4OQOZc4YfY=
+SIZE (blas-1.0/blas.tgz) = 98957
+SIZE (blas-1.0/manpages.tgz) = 1016997
Index: patches/patch-blas_man_manl_cscal_l
===
RCS file: patches/patch-blas_man_manl_cscal_l
diff -N patches/patch-blas_man_manl_cscal_l
--- patches/patch-blas_man_manl_cscal_l 5 Sep 2007 15:32:19 - 1.2
+++ /dev/null 1 Jan 1970 00:00:00 -
@@ -1,9 +0,0 @@
-$OpenBSD: patch-blas_man_manl_cscal_l,v 1.2 2007/09/05 15:32:19 steven Exp $
blas/man/manl/cscal.l.orig Sat Apr 1 19:54:03 2006
-+++ blas/man/manl/cscal.l Sat Apr 1 19:54:11 2006
-@@ -1,4 +1,4 @@
--.SH NAME
-+.TH CSCAL 3f
-
- .SH SYNOPSIS
- .TP 31
Index: patches/patch-blas_man_manl_dcabs1_l
===
RCS file: patches/patch-blas_man_manl_dcabs1_l
diff -N patches/patch-blas_man_manl_dcabs1_l
--- patches/patch-blas_man_manl_dcabs1_l5 Sep 2007 15:32:19 -
1.2
+++ /dev/null 1 Jan 1970 00:00:00 -
@@ -1,9 +0,0 @@
-$OpenBSD: patch-blas_man_manl_dcabs1_l,v 1.2 2007/09/05 15:32:19 steven Exp $
blas/man/manl/dcabs1.l.origSat Apr 1 19:54:21 2006
-+++ blas/man/manl/dcabs1.l Sat Apr 1 19:54:33 2006
-@@ -1,4 +1,4 @@
--.SH NAME
-+.TH DCABS1 3f
-
- .SH SYNOPSIS
- .TP 17
Index: patches/patch-blas_man_manl_zrotg_l
===
RCS file: patches/patch-blas_man_manl_zrotg_l
diff -N patches/patch-blas_man_manl_zrotg_l
--- patches/patch-blas_man_manl_zrotg_l 5 Sep 2007 15:32:19 - 1.2
+++ /dev/null 1 Jan 1970 00:00:00 -
@@ -1,9 +0,0 @@
-$OpenBSD: patch-blas_man_manl_zrotg_l,v 1.2 2007/09/05 15:32:19 steven Exp $
blas/man/manl/zrotg.l.orig Sat Apr 1 19:54:40 2006
-+++ blas/man/manl/zrotg.l Sat Apr 1 19:54:47 2
/ports/math/blas does not build
I have not figured out how to cut and paste from xterm and I expect that attachments are not the thing to do. Thus no nice build outputs. My -current is say two days old. There are four patches and none of the patches cleanly apply. I note it seems to be almost 20 year old code, fortran, and openoffice3 build is dependent on it.
