[SciPy-User] Mismatch architecture type, compiling scipy on Mac 10.6

[Ralf Gommers]

On Thu, Feb 10, 2011 at 8:07 AM, Charles R Harris <charlesr.harris at gmail.com
> wrote:

>
>
> On Wed, Feb 9, 2011 at 4:52 PM, Jeremy Conlin <jlconlin at gmail.com> wrote:
>
>> On Mon, Feb 7, 2011 at 5:32 PM, David <david at silveregg.co.jp> wrote:
>> > On 02/08/2011 07:45 AM, Jeremy Conlin wrote:
>> >> I have compiled python 2.7 and numpy.1.5.1 in 64-bit on my Mac running
>> >> 10.6.3.  I'm trying to compile 64 bit scipy as well, but I'm running
>> >> into a problem.  I'm pretty sure that the problem is architecture
>> >> related.  The error(s) I get are copied below.  The command I used to
>> >> compile scipy is:
>> >>
>> >> LDFLAGS="-arch x86_64 -Wall -undefined dynamic_lookup -bundle"
>> >> FFLAGS="-arch x86_64"  python setup.py install
>> >>
>> >> I used the same command to compile numpy without any problems.  I am
>> >> using g95 and gcc (version info below).
>> >
>> > Don't use g95, it has caused too many issues, especially on mac os x.
>> > Use gfortran, the one available on http://r.research.att.com/tools/(NOT
>> > the one on hpc.sourceforge.org, which does not work well either).
>>
>>
>> Thanks for the suggestion.  Unfortunately I can't install a
>> precompiled binary since I do not have root access on my machine.
>> However, I was able to compile gfortran (version 4.5.2) on my own and
>> it got me past my previous problems.
>>
>> I have now come up with a new error.  It seems that gcc doesn't like
>> the option "-faltivec" (full error copied below).  A Google search
>> indicates that this may be an old option that isn't necessary anymore.
>>  How can I remove this option?  I know I didn't specify it in any
>> environment variable.
>>
>>
> Altivec is a PPC thing, seems strange that it should turn up on Intel.
>
> distutils is probably trying to create fat binaries because gcc supports
that. The recommended binary gfortran does as well, but your home-compiled
one maybe doesn't?

Try the same with numscons, that will probably work better than distutils
(as noted at http://www.scipy.org/Installing_SciPy/Mac_OS_X)

Ralf
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20110210/6360050d/attachment.html>