[SciPy-user] Plea: Installing SciPy on Mac 10.4.9: _fftpack.so problems?

[Nick Fotopoulos]

On 3/17/07, scipy-user-request at scipy.org <scipy-user-request at scipy.org> wrote:
> Message: 3
> Date: Fri, 16 Mar 2007 15:31:35 -0500
> From: Robert Kern <robert.kern at gmail.com>
> Subject: Re: [SciPy-user] Plea: Installing SciPy on Mac 10.4.9:
>         _fftpack.so problems?
> To: SciPy Users List <scipy-user at scipy.org>
> Message-ID: <45FAFEA7.2010403 at gmail.com>
> Content-Type: text/plain; charset=UTF-8

<snip>

> Nick Fotopoulos wrote:
> > I eventually got it to build using "python setup.py build_src
> > build_clib --fcompiler=gnu95 build_ext -lpython --fcompiler=gnu95
> > build", but unsurprisingly, it doesn't work.  I think it's picking up
> > the system python's libpython:
>
> Yes, because you added -lpython which is not how you need to link against the
> Python libraries for framework builds of Python.

I was guessing.  Thanks for the correction.

> > nvf at dirac:~$ python -c "import scipy; scipy.test(10,10)"
> > Fatal Python error: Interpreter not initialized (version mismatch?)
> > Abort trap
> > nvf at dirac:~$ locate libpython
> > /Developer/SDKs/MacOSX10.4u.sdk/usr/lib/libpython.dylib
> > /Developer/SDKs/MacOSX10.4u.sdk/usr/lib/libpython2.3.dylib
> > /Developer/SDKs/MacOSX10.4u.sdk/usr/lib/libpython2.dylib
> > /Library/Frameworks/Python.framework/Versions/2.4/lib/python2.4/config/libpython2.4.a
> > /Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/config/libpython2.5.a
> > /usr/lib/libpython.dylib
> > /usr/lib/libpython2.3.dylib
> > /usr/lib/libpython2.dylib
> >
> > Those libpython.dylibs just simlink to the libpython2.3.dylib.  I
> > would have thought that the MacPython installer would provide dynamic
> > libraries in addition to the static libraries.
>
> No, everything should be contained in /Library/Frameworks/Python.framework . In
> such a case, the "shared library" is actually the file
> /Library/Frameworks/Python.framework/Python . It gets linked when "-framework
> Python" is part of the link command, as distutils does provided it is not
> interfered with.

Huh, I wouldn't have guessed this on my own.

> Can you supply the output from an unsuccessful build again? Don't do -lpython
> this time.

<all the missing _Py symbols>
collect2: ld returned 1 exit status
error: Command "/usr/local/bin/gfortran -Wall -bundle
build/temp.macosx-10.3-fat-2.5/build/src.macosx-10.3-fat-2.5/Lib/fftpack/_fftpackmodule.o
build/temp.macosx-10.3-fat-2.5/Lib/fftpack/src/zfft.o
build/temp.macosx-10.3-fat-2.5/Lib/fftpack/src/drfft.o
build/temp.macosx-10.3-fat-2.5/Lib/fftpack/src/zrfft.o
build/temp.macosx-10.3-fat-2.5/Lib/fftpack/src/zfftnd.o
build/temp.macosx-10.3-fat-2.5/build/src.macosx-10.3-fat-2.5/fortranobject.o
-L/opt/lscsoft/non-lsc/lib
-L/usr/local/lib/gcc/i386-apple-darwin8.8.1/4.3.0
-Lbuild/temp.macosx-10.3-fat-2.5 -ldfftpack -lfftw3 -lgfortran -o
build/lib.macosx-10.3-fat-2.5/scipy/fftpack/_fftpack.so" failed with
exit status 1

Many thanks for the information and for taking a shot at this.

Take care,
Nick