This problem has been documented in https://forums.gentoo.org/viewtopic-t-967644-start-0-postdays-0-postorder-asc-highlight-.html#7377526 ------ I added the science overlay and was now able to start building sci-libs/atlas after (temporarily) unmerging eselect-cblas, eselect-blas, eselect-lapack which were blocking atlas, but then it failed: Code: /usr/bin/gcc -o ATL_dtrsmLLTN.o -c -DL2SIZE=4194304 -I/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/include -I/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//include -I/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//include/contrib -DAdd_ -DF77_INTEGER=int -DStringSunStyle -DATL_OS_Linux -DATL_ARCH_Corei2 -DATL_CPUMHZ=3601 -DATL_AVX -DATL_SSE3 -DATL_SSE2 -DATL_SSE1 -DATL_USE64BITS -DATL_GAS_x8664 -m64 -DWALL -DATL_NCPU=8 -O2 -pipe -ggdb -fPIC -m64 -DDREAL -DLower_ \ -DTranspose_ -DNonUnitDiag_ /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmL.c /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c: In function 'ATL_rk4': /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:31:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:32:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:33:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:34:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:35:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:53:7: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:142:7: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKL_rk4.c:186:7: error: unknown type name '__m256d' make[5]: *** [ATL_dtrsmKL_rk4.o] Error 1 make[5]: *** Waiting for unfinished jobs.... /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c: In function 'ATL_rk4': /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:31:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:32:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:33:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:34:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:35:4: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:53:7: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:142:7: error: unknown type name '__m256d' /var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//src/blas/level3/kernel/ATL_trsmKR_rk4.c:186:7: error: unknown type name '__m256d' make[5]: *** [ATL_dtrsmKR_rk4.o] Error 1 make[5]: Leaving directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/src/blas/level3/kernel' make[4]: *** [dlib] Error 2 make[4]: Leaving directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/src/blas/level3/kernel' make[3]: *** [dkern] Error 2 make[3]: Leaving directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/src/blas/level3' make[2]: *** [IBuildLibs] Error 2 make[2]: Leaving directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/bin' DONE STAGE 5-1-0 at 12:46 ATLAS install complete. Examine ATLAS/bin/<arch>/INSTALL_LOG/SUMMARY.LOG for details. make[1]: Leaving directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared' make clean make[1]: Entering directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared' rm -f *.o x* config?.out *core* make[1]: Leaving directory `/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared' make -j5 libclapack.a make: *** No rule to make target `liblapack.a', needed by `libclapack.a'. Stop. * ERROR: sci-libs/atlas-3.10.1-r1 failed (compile phase): * emake failed * * If you need support, post the output of `emerge --info '=sci-libs/atlas-3.10.1-r1'`, * the complete build log and the output of `emerge -pqv '=sci-libs/atlas-3.10.1-r1'`. * This ebuild is from an overlay named 'science': '/var/lib/layman/science/' * The complete build log is located at '/var/tmp/portage/sci-libs/atlas-3.10.1-r1/temp/build.log'. * The ebuild environment file is located at '/var/tmp/portage/sci-libs/atlas-3.10.1-r1/temp/environment'. * Working directory: '/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/lib' * S: '/var/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS' >>> Failed to emerge sci-libs/atlas-3.10.1-r1, Log file: >>> '/var/tmp/portage/sci-libs/atlas-3.10.1-r1/temp/build.log' There seem to be two problems: - the unknown type name '__m256d' - the attempt to build liblapack.a in order to build libclapack.a. This is especiallz strange since liblapack is present: Code: # ls -l /usr/lib64/liblapack.* lrwxrwxrwx 1 root root 28 Aug 20 09:43 /usr/lib64/liblapack.a -> lapack/reference/liblapack.a lrwxrwxrwx 1 root root 29 Aug 20 09:43 /usr/lib64/liblapack.so -> lapack/reference/liblapack.so lrwxrwxrwx 1 root root 31 Aug 20 09:43 /usr/lib64/liblapack.so.0 -> lapack/reference/liblapack.so.0 and Code: # equery belongs /usr/lib64/liblapack.a * Searching for /usr/lib64/liblapack.a ... sci-libs/lapack-reference-3.1.1-r1 (/usr/lib64/lapack/reference/liblapack.a) At least once, further up in the buyilding of sci-atlas, there was: Code: ERROR IN COMMAND: make mmucaseN mmrout=SSEGENOUTDIR/dgenmm_sse.c moves="" casnam=res/dmmNMK10037_TN60x60x60_2x3x60_0_0x0x0_0_0_a1_b1_0 pre=d M=60 N=60 K=60 mb=60 nb=60 kb=60 mu=2 nu=3 ku=60 lda=60 ldb=60 ldc=0 beta=1 > /dev/null 2>&1 PROPOSED FILENAME: res/dmmNMK10037_TN60x60x60_2x3x60_0_0x0x0_0_0_a1_b1_0 GENSTR='make dgensseK mu=2 nu=3 ku=60 MB=60 NB=60, KB=60 dssegenf=SSEGENOUTDIR/dgenmm_sse.c lda=60 ldb=60 ldc=0 ssegenflags="-treatLoadsAsFloat 1 -treatStoresAsFloat 1 -prefetchABlock 1 -prefetchBCols 1" > /dev/null 2>&1 ' Default compiler & flags do not support SSEGEN!! I don't know what SSEGEN is, but it seems not to be severe because the build continued... --------- Reproducible: Always
$ emerge --info Portage 2.2.7 (default/linux/amd64/13.0/desktop, gcc-4.6.3, glibc-2.15-r3, 3.10.7-gentoo-r1 x86_64) ================================================================= System uname: Linux-3.10.7-gentoo-r1-x86_64-Intel-R-_Core-TM-_i7-2760QM_CPU_@_2.40GHz-with-gentoo-2.2 KiB Mem: 8135224 total, 189904 free KiB Swap: 2047996 total, 2047924 free Timestamp of tree: Thu, 31 Oct 2013 00:45:01 +0000 ld GNU ld (GNU Binutils) 2.23.1 ccache version 3.1.9 [enabled] app-shells/bash: 4.2_p45 dev-java/java-config: 2.1.12-r1 dev-lang/python: 2.7.5-r3, 3.2.5-r3 dev-util/ccache: 3.1.9 dev-util/cmake: 2.8.11.2 dev-util/pkgconfig: 0.28 sys-apps/baselayout: 2.2 sys-apps/openrc: 0.11.8 sys-apps/sandbox: 2.6-r1 sys-devel/autoconf: 2.13, 2.69 sys-devel/automake: 1.11.6, 1.12.6, 1.13.4 sys-devel/binutils: 2.23.1 sys-devel/gcc: 4.6.3, 4.7.3-r1 sys-devel/gcc-config: 1.7.3 sys-devel/libtool: 2.4.2 sys-devel/make: 3.82-r4 sys-kernel/linux-headers: 3.9 (virtual/os-headers) sys-libs/glibc: 2.15-r3 Repositories: gentoo science ACCEPT_KEYWORDS="amd64" ACCEPT_LICENSE="* -@EULA" CBUILD="x86_64-pc-linux-gnu" CFLAGS="-Os -pipe -ggdb" CHOST="x86_64-pc-linux-gnu" CONFIG_PROTECT="/etc /usr/share/config /usr/share/gnupg/qualified.txt /usr/share/maven-bin-3.0/conf /usr/share/polkit-1/actions /var/lib/hsqldb" CONFIG_PROTECT_MASK="/etc/ca-certificates.conf /etc/dconf /etc/env.d /etc/fonts/fonts.conf /etc/gconf /etc/gentoo-release /etc/php/apache2-php5.4/ext-active/ /etc/php/cgi-php5.4/ext-active/ /etc/php/cli-php5.4/ext-active/ /etc/revdep-rebuild /etc/sandbox.d /etc/terminfo /etc/texmf/language.dat.d /etc/texmf/language.def.d /etc/texmf/updmap.d /etc/texmf/web2c" CXXFLAGS="-Os -pipe -ggdb" DISTDIR="/mnt/data/tmp/portage/distfiles/" FCFLAGS="-O2 -pipe" FEATURES="assume-digests binpkg-logs ccache config-protect-if-modified distlocks ebuild-locks fixlafiles merge-sync news nodoc parallel-fetch preserve-libs protect-owned sandbox sfperms splitdebug strict unknown-features-warn unmerge-logs unmerge-orphans userfetch userpriv usersandbox usersync" FFLAGS="-O2 -pipe" GENTOO_MIRRORS="http://distfiles.gentoo.org" LDFLAGS="-Wl,-O1 -Wl,--as-needed" MAKEOPTS="-j8" PKGDIR="/mnt/data/tmp/portage/packages/" PORTAGE_CONFIGROOT="/" PORTAGE_RSYNC_OPTS="--recursive --links --safe-links --perms --times --omit-dir-times --compress --force --whole-file --delete --stats --human-readable --timeout=180 --exclude=/distfiles --exclude=/local --exclude=/packages" PORTAGE_TMPDIR="/mnt/data/tmp/" PORTDIR="/mnt/portage" PORTDIR_OVERLAY="/var/lib/layman/science" SYNC="rsync://rsync.de.gentoo.org/gentoo-portage" USE="3dnow 3dnowext X a52 aac acl acpi alsa amd64 apng avahi berkdb bitmap-fonts bluetooth branding bzip2 cairo cdda cdr cli consolekit cracklib crypt cups cxx dbus djvu dri dts dvd dvdr eds emboss encode evo exif fam firefox flac fontconfig fortran gdbm gif gnome gpm gstreamer gtk iconv icu ipv6 jpeg kde lcms ldap libass libnotify mad matroska mmx mmxext mng modules mp3 mp4 mpeg mudflap multilib nautilus ncurses networkmanager nls nptl ogg ogv opengl openmp pam pango pcre pdf png policykit ppds projectm qt3support qt4 readline samba sdl semantic-desktop session spell sse sse2 sse4_1 sse4_2 ssl ssse3 startup-notification svg tcpd theora threads tiff truetype truetype-fonts type1-fonts udev udisks unicode upnp upower usb vorbis vpx webkit webm widescreen wifi win32codecs wma wxwidgets x264 xcb xfce xine xinerama xml xscreensaver xulrunner xv xvid xz zeroconf zlib" ABI_X86="64" ALSA_CARDS="ali5451 als4000 atiixp atiixp-modem bt87x ca0106 cmipci emu10k1x ens1370 ens1371 es1938 es1968 fm801 hda-intel intel8x0 intel8x0m maestro3 trident usb-audio via82xx via82xx-modem ymfpci" APACHE2_MODULES="authn_core authz_core socache_shmcb unixd actions alias auth_basic authn_alias authn_anon authn_dbm authn_default authn_file authz_dbm authz_default authz_groupfile authz_host authz_owner authz_user autoindex cache cgi cgid dav dav_fs dav_lock deflate dir disk_cache env expires ext_filter file_cache filter headers include info log_config logio mem_cache mime mime_magic negotiation rewrite setenvif speling status unique_id userdir usertrack vhost_alias" CALLIGRA_FEATURES="kexi words flow plan sheets stage tables krita karbon braindump author" CAMERAS="ptp2" COLLECTD_PLUGINS="df interface irq load memory rrdtool swap syslog" ELIBC="glibc" GPSD_PROTOCOLS="ashtech aivdm earthmate evermore fv18 garmin garmintxt gpsclock itrax mtk3301 nmea ntrip navcom oceanserver oldstyle oncore rtcm104v2 rtcm104v3 sirf superstar2 timing tsip tripmate tnt ubx" INPUT_DEVICES="evdev synaptics" KERNEL="linux" LCD_DEVICES="bayrad cfontz cfontz633 glk hd44780 lb216 lcdm001 mtxorb ncurses text" LIBREOFFICE_EXTENSIONS="presenter-console presenter-minimizer" LINGUAS="de en" OFFICE_IMPLEMENTATION="libreoffice" PHP_TARGETS="php5-5" PYTHON_SINGLE_TARGET="python2_7" PYTHON_TARGETS="python2_7 python3_2" RUBY_TARGETS="ruby19 ruby18" USERLAND="GNU" VIDEO_CARDS="evdev synaptics nouveau" XTABLES_ADDONS="quota2 psd pknock lscan length2 ipv4options ipset ipp2p iface geoip fuzzy condition tee tarpit sysrq steal rawnat logmark ipmark dhcpmac delude chaos account" Unset: CPPFLAGS, CTARGET, EMERGE_DEFAULT_OPTS, INSTALL_MASK, LANG, LC_ALL, PORTAGE_BUNZIP2_COMMAND, PORTAGE_COMPRESS, PORTAGE_COMPRESS_FLAGS, PORTAGE_RSYNC_EXTRA_OPTS, USE_PYTHON
This upstream bug is possibly connected http://sourceforge.net/p/math-atlas/support-requests/842/
This is the first failure on my machine: make[5]: Entering directory `/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/tune/blas/gemm' /usr/lib64/ccache/bin/gcc -DL2SIZE=4194304 -I/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/include -I/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//include -I/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//include/contrib -DAdd_ -DF77_INTEGER=int -DStringSunStyle -DATL_OS_Linux -DATL_ARCH_Corei2 -DATL_CPUMHZ=2401 -DATL_AVX -DATL_SSE3 -DATL_SSE2 -DATL_SSE1 -DATL_USE64BITS -DATL_GAS_x8664 -m64 -DWALL -DATL_NCPU=8 -Os -pipe -ggdb -fPIC -m64 -o xmmksearch_sse mmksearch_sse.o ./xmmksearch_sse -p z 65 of /mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS//include/atlas_mmtesttime.h: FAILED COMMAND : make cmmutstcase mmrout=zgenmm_sse.c csC=2 pre=z M=48 N=48 K=48 mb=48 nb=48 kb=48 lda=48 ldb=48 ldc=100 > /dev/null 2>&1 genstr was = 'make zgensseK mu=1 nu=1 ku=1 MB=48 NB=48, KB=48 zssegenf=zgenmm_sse.c lda=48 ldb=48 ldc=0 ssegenflags="" > /dev/null 2>&1 ' ERROR IN COMMAND: make cmmucase mmrout=SSEGENOUTDIR/zgenmm_sse.c csC=2 moves="" casnam=res/zmmNMK10037_TN54x54x54_2x3x54_0_0x0x0_0_0_a1_b1_0 pre=z M=54 N=54 K=54 mb=54 nb=54 kb=54 mu=2 nu=3 ku=54 lda=54 ldb=54 ldc=0 beta=1 > /dev/null 2>&1 PROPOSED FILENAME: res/zmmNMK10037_TN54x54x54_2x3x54_0_0x0x0_0_0_a1_b1_0 GENSTR='make zgensseK mu=2 nu=3 ku=54 MB=54 NB=54, KB=54 zssegenf=SSEGENOUTDIR/zgenmm_sse.c lda=54 ldb=54 ldc=0 ssegenflags="-treatLoadsAsFloat 1 -treatStoresAsFloat 1 -prefetchABlock 1 -prefetchBCols 1" > /dev/null 2>&1 ' Default compiler & flags do not support SSEGEN!! make[5]: *** [res/zMMKSSE.sum] Error 255 make[5]: Leaving directory `/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/tune/blas/gemm' EXTERNAL SEARCH FAILED: make res/zMMKSSE.sum
The upstream bug refers to invalid build flags. The ones used by this build (written out by ebuild to .../portage/sci-libs/atlas-3.10.1-r1/work/ATLAS_shared/myconf.out) are: --prefix=/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/image//usr --libdir=/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/image//usr/lib64 --incdir=/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/image//usr/include --cc=x86_64-pc-linux-gnu-gcc -D c -DWALL -C acg 'gcc' -F acg '-Os -pipe -g' -Ss pmake '$(MAKE) -j8' -t -1 -Si omp 0 -b 64 -C if '/usr/bin/x86_64-pc-linux-gnu-gfortran' -F if '-O2 -pipe' -Si latune 1 --with-netlib-lapack-tarfile=/mnt/data/tmp/portage/sci-libs/atlas-3.10.1-r1/distdir/lapack-3.4.2.tgz -Fa alg -fPIC
I can confirm that the problem is not related to the CFLAGS or ccache. I disabled the CFLAGS and CXXFLAGS variables in make.conf and the ccache FEATURE, but the same problem appears.
Created attachment 366042 [details] /proc/cpuinfo
Hi, first you should migrate properly to blas/lapack from the science overlay. I recommend following this document: https://wiki.gentoo.org/wiki/User_talk:Houseofsuns Lapack from atlas is separate from any other lapack you have already installed so it doesn't matter you already have one copy installed. Once you are all properly set up we'll lock at that error if it still happens.
(In reply to Francois Bissey from comment #7) > Hi, > first you should migrate properly to blas/lapack from the science overlay. > I recommend following this document: > https://wiki.gentoo.org/wiki/User_talk:Houseofsuns > Lapack from atlas is separate from any other lapack you have already > installed so it doesn't matter you already have one copy installed. > > Once you are all properly set up we'll lock at that error if it still > happens. Hi Francois, thank you for getting back to me. I followed the document, and now have: sci-libs/blas-reference 20131116[1] sci-libs/cblas-reference 20110218[1] virtual/blas 2.1[1] virtual/cblas 2.0[1] app-admin/eselect 1.4-r100[1] sci-libs/atlas 3.10.1-r1[1] [1] "science" /var/lib/layman/science But nothing has changed. However, in the meantime I was able to install atlas with the "generic" use flag. Using "-static-libs" did not help (I thought it might then not build libclapack.a). Now that is not a good solution as a generic build will be relatively slow.
Relatively. You need to attach the build log so we can figure out what configure options were passed and how it is trying to build things. portage should tell you where it is located when the build fails. Compress it and attach it please. Do you really need *atlas* in particular? openblas offers quite good performances for a number of cpus by using hand tuned assembly. Of course lapack itself is not optimised but even with atlas the optimisation of lapack is only for a handful of functions.
Actually google is very helpful here: http://trac.macports.org/ticket/38816 So this particular error should disappear if you add "-mavx" to your CFLAGS to compile atlas. It is tricky to try to add logic to spot this in the ebuild. Possibly we have to filter flags...
Created attachment 366128 [details] sci-libs/atlas-3.10.1-r1 build log
> Do you really need *atlas* in particular? openblas offers quite good > performances for a number of cpus by using hand tuned assembly. Of course > lapack itself is not optimised but even with atlas the optimisation of > lapack is only for a handful of functions. I am not sure. I need numpy/scipy and octave which use the lapack/blas bunch. (In reply to Francois Bissey from comment #10) > http://trac.macports.org/ticket/38816 > So this particular error should disappear if you add "-mavx" to your CFLAGS > to compile atlas. This worked for me. Thank you for digging this up! > It is tricky to try to add logic to spot this in the ebuild. Possibly we > have to filter flags... I also don't understand what is particular about my env that this problem triggered here, and apparently works ok for others (assuming I'm not the only person in the world using gentoo + atlas).
numpy/scipy only need virtual/blas virtual/lapack which in turn will make sure you have a blas and lapack on your system. You need to be sure to eselect a blas and lapack before merging anything that depends on them otherwise we'll get interesting build logs. Using atlas for lapack in scipy create a few extra direct interfaces to clapack that won't be there with lapack-reference but I don't believe that's visible at the user level (unless you are an expert) it's all under the hood. Issues about atlas crop every so often. It is a tricky package ;)
Now I have all these packages installed, but numpy can not load because the symlinks /usr/lib/libcblas.so /usr/lib/libblas.so /usr/lib/liblapack.so are orphaned! eselect does not handle them, it only does the pc files (e.g. /usr/lib64/pkgconfig/lapack.pc): eselect blas set atlas eselect cblas set atlas eselect lapack set atlas Using "eselect ... files" I find only these files are managed: /usr/lib64/pkgconfig/blas.pc /usr/lib64/pkgconfig/cblas.pc /usr/include/cblas.h /usr/lib64/pkgconfig/lapack.pc I had to symlink them manually sudo ln -s /usr/lib64/libatllapack.so.3 /usr/lib/liblapack.so sudo ln -s /usr/lib64/libatllapack.so.3 /usr/lib/liblapack.so.0 sudo ln -s /usr/lib/libatlcblas.so /usr/lib/libcblas.so sudo ln -s /usr/lib/libopenblas.so /usr/lib/libblas.so Now "python -c 'import numpy'" works again. But this is a really bad idea. How is this supposed to work since eselect does not manage these symlinks?
(In reply to Johannes Buchner from comment #14) > Now I have all these packages installed, but numpy can not load because the > symlinks > > /usr/lib/libcblas.so > /usr/lib/libblas.so > /usr/lib/liblapack.so > > are orphaned! They unfortunately have to be manually removed. Do not keep them, do not replace them or things will break in "interesting ways". > eselect does not handle them, it only does the pc files (e.g. > /usr/lib64/pkgconfig/lapack.pc): > eselect blas set atlas > eselect cblas set atlas > eselect lapack set atlas > > Using "eselect ... files" I find only these files are managed: > /usr/lib64/pkgconfig/blas.pc > /usr/lib64/pkgconfig/cblas.pc > /usr/include/cblas.h > /usr/lib64/pkgconfig/lapack.pc > > Correct. All ebuilds that have a dependency to blas/lapack are supposed to use the .pc files. If you find one that doesn't fill a bug here straight away. > I had to symlink them manually > sudo ln -s /usr/lib64/libatllapack.so.3 /usr/lib/liblapack.so > sudo ln -s /usr/lib64/libatllapack.so.3 /usr/lib/liblapack.so.0 > sudo ln -s /usr/lib/libatlcblas.so /usr/lib/libcblas.so > sudo ln -s /usr/lib/libopenblas.so /usr/lib/libblas.so > See above don't do that. > Now "python -c 'import numpy'" works again. But this is a really bad idea. > How is this supposed to work since eselect does not manage these symlinks? Once you have updated your blas/lapack stack all the packages using blas/lapack have to be rebuild to be linked to the correct library files. So once you have removed all these links, proceed to emerge numpy/scipy/octave again.
Works ok now. The change I had not understood was that instead of runtime linking, the packages are now linked to the specific libraries at compile time.
