[SciPy-user] Trying to build scipy 32-bit on a 64-bit machine

[Jeremy Mayes]

Hi,

Thanks for the response.  It wasn't too bad building python.  And let me
change my initial statement slightly.  I'm trying to build 32-bit binaries
to run on an x86_64 machine.  You only have to pass gcc the -m32 flag to get
it to do that, so, it's not horrible.  Most configure scripts will pay
attention to the CFLAGS env variable.  It's when I get into distutils land
that I'm less sure of what to do.  It definitely seems that distutils also
pays attention to the CFLAGS/LDFLAGS env variables, but, not sure about
numpy/scipy for fortran.  I've tried FFLAGS, but, that leads to other
problems in addition to the numerous threads I've read on NOT setting those
vars when building.

On Tue, Feb 26, 2008 at 6:01 AM, David Cournapeau <
david at ar.media.kyoto-u.ac.jp> wrote:

> Jeremy Mayes wrote:
> > Hi,
> >
> > I'm trying to build scipy for a target of i686 ( i.e., 32-bit ) but on
> > an x86_64 host using gcc/4.1.1.  I've been struggling with this and
> > haven't seen any reference in the archives ( I apologize if I missed it
> ).
>
> It won't be easy: distutils (the python package used to build numpy)
> does not support cross-compiling. Already cross-compiling python itself
> is difficult, and you will need that first.
>
> >
> > I've been trying to set CFLAGS and LDFLAGS to pass -m32, but, I get
> > undefined symbol errors ( MAIN__ ).  If I just let it run, then, I get
> > errors with LONG_BIT defined in pyport.h ( python successfully build
> > 32-bit ).
>
> Modifying flags will not work. Different architectures have different
> python installations (different headers, with different values: that's
> certainly the cause of the above error). Do you have any experience
> cross-compiling ? Because cross-compilation is already difficult, and
> python is not an easy package to cross-compile (bootstrapping issues,
> etc...), specially since the installation process of python does not
> support cross-compilation (you can find patches, but I don't know if
> they are updated for recent python).
>
> cheers,
>
> David
> _______________________________________________
> SciPy-user mailing list
> SciPy-user at scipy.org
> http://projects.scipy.org/mailman/listinfo/scipy-user
>



-- 
--jlm
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mail.scipy.org/pipermail/scipy-user/attachments/20080226/5e21347f/attachment.html>