[SciPy-user] building scipy from source on Mac Os X 10.4 ppc

Robert Kern robert.kern at gmail.com
Fri Aug 17 15:32:11 EDT 2007


Samuel M. Smith wrote:
> Zach, Robert
> 
> Changing to the other fortran compiler did the trick. Yeah! Thank you!
> Still have to explicitly set the MACOSX_DEPLOYMENT_TARGET to 10.4 or  
> else it builds by default for 10.3

You should leave off the MACOSX_DEPLOYMENT_TARGET unless if you really need to
use a feature that is only available in 10.4.

> I don't know who the maintainer of the scipy web page for os x  
> installation is but I suggest it be updated
> to use a fortran compiler that works.  Also the link for the g77  
> compiler is broken it points to the g95
> compiler both on the hpc site.

It's a wiki, so the set of maintainers contains you, too.  :-)  In particular,
people who have found problems and solved them are often the best people to
write documentation about how to avoid such problems. If you have the time, we
would appreciate your fixes to that page.

> ***************
> In case anyone is interested here is what I did to install scipy from  
> source on a G4 powerbook
> os x 10.4.10
> xcode 2.4.1     xcode_2.4.1_8m1910_6936315.dmg
> gcc 4.01
> 
> Python and numpy from http://www.pythonmac.org/packages/py25-fat/ 
> index.html
> python-2.5.1-macosx.dmg
> wxPython2.8-osx-unicode-2.8.3.0-universal10.4-py2.5.dmg
> pytz-2006g-py2.5-macosx10.4.dmg
> numpy-1.0.3-py2.5-macosx10.4.mpkg
> 
> $ echo $PATH
> /usr/local/bin:/usr/local/sbin:/usr/texbin:/Library/Frameworks/ 
> Python.framework/Versions/Current/bin:/opt/local/bin:/opt/local/sbin:/ 
> bin:/sbin:/usr/bin:/usr/sbin:/Users/samuel/bin
> 
> 
> *** check gcc version
> $ gcc_select
> Current default compiler:
> gcc version 4.0.1 (Apple Computer, Inc. build 5367)
> 
> to change if not 4.0x
> $ sudo gcc_select 4.0
> 
> *** Install subversion 1.4.4-2  from pkg installer to get scipy from svn
> http://www.open.collab.net/servlets/OCNDownload?id=CSVNMACC
> Subversion 1.4.4-2 Universal.dmg
> 
> **** Install gfortran
> get gfortran from below instead of links on scipy page
> http://r.research.att.com/tools/
> http://r.research.att.com/gfortran-4.2.1.dmg
> 
> *** Install fftw
> get fftw
> http://fftw.org/fftw-3.1.2.tar.gz
> $ tar -xvzf fftw-3.1.2.tar.gz
> $ cd fftw-3.1.2
> $ ./configure
> $ make
> $ sudo make install
> $ sudo  ln -s /usr/local/lib/libfftw3.a /usr/local/lib/libfftw.a
> $ sudo  ln -s /usr/local/lib/libfftw3.la /usr/local/lib/libfftw.la
> $ sudo  ln -s /usr/local/include/fftw3.h /usr/local/include/fftw.h
> 
> *** Build and Install scipy from svn
> $ cd /Volumes/Archive/Install/Python/MacPython/Python2.5.x/scipy/
> $ svn co http://svn.scipy.org/svn/scipy/trunk scipysvn
> Checked out revision 3245.
> 
> $ cd scipysvn
> 
> ***must set environment variable or will build for 10.3 not 10.4
> $ export MACOSX_DEPLOYMENT_TARGET=10.4
> 
> $ python setup.py build_src build_clib --fcompiler=gnu95 build_ext -- 
> fcompiler=gnu95 build
> $sudo python setup.py install
> 
> *** to test
> $ python
> 
>  >>> import scipy
>  >>> scipy.test(1,10)
> 
> I had 3 failures
> 
> ...
> 
> ****************************************************************
> WARNING: clapack module is empty
> -----------
> See scipy/INSTALL.txt for troubleshooting.
> Notes:
> * If atlas library is not found by numpy/distutils/system_info.py,
>    then scipy uses flapack instead of clapack.
> ****************************************************************
> .....

I may have fixed that in numpy SVN; I'm not sure.

The check_dot failures are known and we have a fix thanks to David Cournapeau
that will probably make its way in soon.

-- 
Robert Kern

"I have come to believe that the whole world is an enigma, a harmless enigma
 that is made terrible by our own mad attempt to interpret it as though it had
 an underlying truth."
  -- Umberto Eco



More information about the SciPy-User mailing list