Glad you sorted it out!
In principle I think it should be possible to run with the standard numpy as well.
At least you are up and running now, which is the most important thing :)
Andy

On 10/18/2012 12:23 AM, Andrew Godbehere wrote:
Problem solved.

Brief: sklearn depends on builds of numpy and scipy that link to atlas. On MacOSX, default builds of numpy and scipy through macports link to the Accelerate framework, and not to atlas. After forcing the atlas variants of numpy and scipy, sklearn works fine.

Action: The implication here is that there is a bug in the macports port file. Specific variants should be listed as dependencies rather than simply "py27-numpy" and "py27-scipy" in the case of Python 2.7.

Thanks everyone for your help!
Best,
Andrew

On Oct 17, 2012, at 3:53 PM, Andrew Godbehere <[email protected] <mailto:[email protected]>> wrote:

How exactly would I link this manually? My sense is that I'd need to edit the setup.py script. Is there another way?

Thanks,
Andrew



On Oct 17, 2012, at 2:01 PM, [email protected] <mailto:[email protected]> wrote:

can you try linking against libatlas manually? that should do it. then i'll to fix the setup.py <http://setup.py/>



Andrew Godbehere <[email protected] <mailto:[email protected]>> schrieb:

    Hi Andy,

    I found _ATL_drotg defined in /opt/local/lib/libatlas.a.
    _ATL_drotg is listed as an undefined reference in libcblas.a,
    libf77blas.a, and libptcblas.a.

    That said, I'm mystified as to why these symbols cannot be
    located. Does this hint at a potential problem in linking?

    Thanks again,
    Andrew

    On Oct 16, 2012, at 3:26 PM, Andrew Godbehere
    <[email protected]
    <mailto:[email protected]>> wrote:

    Also, for some reason, scikit-learn tries to build for posix on
    my system, rather than mac osx.

    On Oct 16, 2012, at 3:07 PM, Andrew Godbehere
    <[email protected]
    <mailto:[email protected]>> wrote:

    Hi Andy,

    I was suspicious of clang, I'd rather this build with gcc.
    I'll look into changing that.


    Here's the output from distutils on blas_opt:
    {'extra_link_args': ['-Wl,-framework', '-Wl,Accelerate'],
    'extra_compile_args': ['-msse3',
    '-I/System/Library/Frameworks/vecLib.framework/Headers'],
    'define_macros': [('NO_ATLAS_INFO', 3)]}

    Do you see anything suspicious there?

    I'm trying to find _ATL_drotg, but I'm first having trouble
    locating the appropriate .so files on my system. I'll have
    more info once I track them down. I'll check cblas and atlas.

    Thanks for the help!

    Best,
    Andrew

    On Oct 16, 2012, at 1:11 PM, Andreas Mueller
    <[email protected] <mailto:[email protected]>>
    wrote:

    Hi Andrew.
    Let's see if we can get to the bottom of this.
    I'm really no expert here.

    On 10/16/2012 09:01 PM, Andrew Godbehere wrote:
    otools (as opposed to ldd) on arrayfuncs.so yields:
    (1) /usr/lib/libSystem.B.dylib (compatibility version 1.0.0,
    current version 169.3.0)
    (2) /System/Library/Frameworks/Accelerate.framework/Versions/A/Accelerate
    (compatibility version 1.0.0, current version 4.0.0)

    I'm hoping this isn't an issue with OSX Accelerate.

    arrayfuncs.so was built as follows:

    /usr/bin/clang -bundle -undefined dynamic_lookup
    -L/opt/local/lib
    build/temp.macosx-10.8-x86_64-2.7/sklearn/utils/arrayfuncs.o
    -Lbuild/temp.macosx-10.8-x86_64-2.7 -lcblas -lm -o
    sklearn/utils/arrayfuncs.so -Wl,-framework -Wl,Accelerate

    Err. Has any one tried building with clang yet? But I guess
    that's not the problem ;)
    I would have expected some atlas linker flags.
    Do you by any chance know where
    _ATL_drotg is defined? I have drotg from cblas and it is
    defined in cblas.so. So I would guess some linker flag is
    missing.

    Also, could you please run:

    |from numpy.distutils.system_info import get_info; print
    get_info('blas_opt', 0)|

    This should tell us what numpy things are good compiler flags.

    Andy
    
------------------------------------------------------------------------------
    Don't let slow site performance ruin your business. Deploy
    New Relic APM
    Deploy New Relic app performance management and know exactly
    what is happening inside your Ruby, Python, PHP, Java, and
    .NET app
    Try New Relic at no cost today and get our sweet Data Nerd
    shirt too!
    
http://p.sf.net/sfu/newrelic-dev2dev_______________________________________________
    Scikit-learn-general mailing list
    [email protected]
    <mailto:[email protected]>
    https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

    
------------------------------------------------------------------------------
    Everyone hates slow websites. So do we.
    Make your web apps faster with AppDynamics
    Download AppDynamics Lite for free today:
    
http://p.sf.net/sfu/appdyn_sfd2d_oct_______________________________________________
    Scikit-learn-general mailing list
    [email protected]
    <mailto:[email protected]>
    https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

    
------------------------------------------------------------------------------
    Everyone hates slow websites. So do we.
    Make your web apps faster with AppDynamics
    Download AppDynamics Lite for free today:
    
http://p.sf.net/sfu/appdyn_sfd2d_oct_______________________________________________
    Scikit-learn-general mailing list
    [email protected]
    <mailto:[email protected]>
    https://lists.sourceforge.net/lists/listinfo/scikit-learn-general


    ------------------------------------------------------------------------

    Everyone hates slow websites. So do we.
    Make your web apps faster with AppDynamics
    Download AppDynamics Lite for free today:
    http://p.sf.net/sfu/appdyn_sfd2d_oct


    ------------------------------------------------------------------------

    Scikit-learn-general mailing list
    [email protected]  
<mailto:[email protected]>
    https://lists.sourceforge.net/lists/listinfo/scikit-learn-general


--
Diese Nachricht wurde von meinem Android-Mobiltelefon mit K-9 Mail gesendet.
------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct_______________________________________________
Scikit-learn-general mailing list
[email protected] <mailto:[email protected]>
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general



------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct


_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

------------------------------------------------------------------------------
Everyone hates slow websites. So do we.
Make your web apps faster with AppDynamics
Download AppDynamics Lite for free today:
http://p.sf.net/sfu/appdyn_sfd2d_oct
_______________________________________________
Scikit-learn-general mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/scikit-learn-general

Reply via email to