[SciPy-User] Problem installing scipy on 10.6.4, python 2.7

Ralf Gommers ralf.gommers@googlemail....
Tue Oct 19 09:16:57 CDT 2010


On Mon, Oct 18, 2010 at 11:10 PM, Benjamin Buch <benni.buch@gmx.de> wrote:
> Hi Ralf,
>
> thanks for your answer!
>
> Am 18.10.2010 um 14:48 schrieb Ralf Gommers:
>
>> You don't seem to be replying to another email, so I'm not sure
>> exactly what you did. Can you please give us:
>
> I was replying to this email directly on gmane:
> http://permalink.gmane.org/gmane.comp.python.scientific.user/26559
> However, it seems like I somehow messed up the replying process..
>
> As I deleted all I've done yesterday, I gave it a new try today - which yielded different results
> which I will describe here.

No problem. There is indeed still a problem, #1399 is reopened. You
can do two things. One is to wait a few days. The other one is to use
numpy 1.5.0 (without the patches you mentioned), and remove "ppc64"
from line 257 of numpy/distutils/fcompiler/gnu.py by hand. Sorry for
the trouble.


>
>> 1. exact numpy/scipy versions you used (version or git commit nr)
>
> numpy:
>>>> numpy.__version__
> '1.5.0'
>
> scipy:
> Recent svn checkout: revision 6845.
>
>> 2. what patches you used
>
> http://projects.scipy.org/numpy/ticket/1399 - this was already patched in my numpy install.
> http://github.com/matthew-brett/numpy/compare/numpy:master...farchs-from-c
>
>> 3. the build command you used and the complete build log
>
> It now builds and installs with
> python setup.py build
> python setup.py install
>
> But I did 'export MACOSX_DEPLOYMENT_TARGET=10.6' before - read it here:
> http://mail.scipy.org/pipermail/numpy-discussion/2010-August/052227.html
> This is the only thing I did differently than yesterday (as far as I can remember).
>
> However, if I now do
>>>> import scipy
>>>> scipy.test()
>
> I get some errors.
> See this gist:
> http://gist.github.com/632345
>
>
>> 4. gfortran details (see below)
>>
>>>
>>> Traceback (most recent call last):
>>> <snip>
>>> File "/Library/Frameworks/Python.framework/
>>> Versions/2.7/lib/python2.7/site-
>>> packages/numpy/distutils/fcompiler/
>>> __init__.py", line 812, in new_fcompiler
>>>    load_all_fcompiler_classes()
>>>  File "/Library/Frameworks/Python.framework/
>>> Versions/2.7/lib/python2.7/site-
>>> packages/numpy/distutils/fcompiler/
>>> __init__.py", line 734, in load_all_fcompiler_classes
>>>    fcompiler_aliases[alias][1].__name__))
>>> ValueError: alias 'g77' defined for both
>>> GnuFCompiler and GnuFCompiler
>>>
>>>
>>> So it seems like I would need the g77 compiler instead of gfortran?
>>> Or is there a setting on my machine wrong?
>>> I'm pretty noob to all this compiling stuff so I don't know what to do...
>>>
>>> I'm on OSX 10.6.4, use python.org 32-bit python and gfortran 4.2.1.
>>
>> The recommended gfortran is from http://r.research.att.com/tools/, the
>> version should be 4.2.3. Where did you get yours?
>
> I installed gfortran via homebrew:
> http://mxcl.github.com/homebrew/
>
> I think it should be 4.2.3, as 'brew info gfortran' gives me

Never seen or heard of brew. To be sure I'd just grab the recommended
binary installer anway, but it is not the issue you're seeing.

Cheers,
Ralf


>
> $ brew info gfortran
> gfortran 4.2.4-5664
> http://r.research.att.com/tools/
> /usr/local/Cellar/gfortran/4.2.4-5664 (28 files, 83M)
>
> But gfortran itself tells me otherwise:
> $ gfortran --version
> GNU Fortran (GCC) 4.2.1 (Apple Inc. build 5664)
> Copyright (C) 2007 Free Software Foundation, Inc.
>
>> Cheers,
>> Ralf
>
> Best,
> Benjamin
>
> _______________________________________________
> SciPy-User mailing list
> SciPy-User@scipy.org
> http://mail.scipy.org/mailman/listinfo/scipy-user
>


More information about the SciPy-User mailing list