[SciPy-dev] {Num, Sci}Py on 64bit Intel Macs with icc, ifort, mkl

[Ulrik Günther]

Hi David,

first, thanks for your fast reply.
> Yes, normally you can disable one lib with an environment variable  
> (such
> as MKL=None will disable MKL even if detected), but this is not
> implemented for the Accelerate framework, from quickly looking at the
> sources.
Yeah, noticed that also. Accelerate is checked first with no option to  
disable it :/

> Solving the Accelerate framework problem should be relatively easy:  
> look
> at numpy/distutils/system_info.py.  The link problems, I don't know,
> since you have not given enough details for them.
The problem is, when compiling/linking, for example, multiarray.so,  
the compiler spits
out many not found symbols, primarily due to not appending -framework  
Python to the
compiling options. The problem can be solved be manually compiling the  
file, while appending
the respective option (along with -undefined dynamic_lookup, as _main  
is not found else, which
may also be a Fortran-compiler related problem, so maybe -no-formain  
for the ifort is more
appropriate than using dynamic lookups...)

> The numpy build
> process is based on the python package distutils: distutils is a  
> really
> badly designed piece of code,
I noticed that on several occasions ;)

> and extremely fragile to changes (it is
> one more proof that you can write Fortran in any language) and jumping
> in is not easy.
>
> I personally would be interested in supporting this configuration with
> my own project numscons, which is a tentative to circumvent distutils
> limitations to build numpy and scipy.
I already looked into the SCons version, but I was not sure about its  
state in the project, as
there is no really information on the website. In fact, my currently  
running version of 64bit
{Num,Sci}py was built using the SCons branch, with a bit of patching  
around.

Could you please tell me what I need now for using the SCons branch?  
Updating my old version
via svn up didn't work, so the first thing I thought was that it has  
been discontinued. I'm very
glad to hear that this is not the case. If I manage to get a working  
build on my configuration,
may I direct to patches to you?

Thanks,

ulrik
-------------- next part --------------
A non-text attachment was scrubbed...
Name: PGP.sig
Type: application/pgp-signature
Size: 194 bytes
Desc: This is a digitally signed message part
URL: <http://mail.python.org/pipermail/scipy-dev/attachments/20080611/f0ff4353/attachment.sig>