[SciPy-user] Scipy.test() fails on Linux 64

Nils Wagner nwagner at iam.uni-stuttgart.de
Fri Feb 15 12:12:04 EST 2008


On Fri, 15 Feb 2008 17:47:06 +0100
  "BERTHE Loic" <berthe.loic at gmail.com> wrote:
>  Hi,
> 
> I'm trying to install scipy from source on a linux 64 
>box.
> Here is my configuration :
> > uname -a
> Linux 2.4.21-32.ELsmp #1 SMP Fri Apr 15 21:03:28 EDT 
>2005 x86_64 x86_64
> x86_64 GNU/Linux
> > gcc --version
> gcc (GCC) 3.2.3 20030502 (Red Hat Linux 3.2.3-52)
> 
> I've installed python 2.5.1, lapack 3.1.1, atlas 3.8, 
>numpy 1.0.4.
> I've seen no compilation errors, and Numpy's test suite 
> runs OK.
> 
> Then, I tried to install scipy 0.60 .
> It compiles fine, but I've got errors when running the 
>test suite :
> 
>> grep FAIL 11_scipy.bash.log
> Warning: FAILURE importing tests for <module 
>'scipy.linsolve.umfpack' from
> '.../linsolve/umfpack/__init__.pyc'>
> Warning: FAILURE importing tests for <module 
>'scipy.linsolve.umfpack.umfpack'
> from '...y/linsolve/umfpack/umfpack.pyc'>
>FAIL: check_syevr 
>(scipy.lib.tests.test_lapack.test_flapack_float)
>FAIL: check_syevr_irange 
>(scipy.lib.tests.test_lapack.test_flapack_float)
>FAIL: check_simple 
>(scipy.linalg.tests.test_decomp.test_eig)
>FAIL: check_simple 
>(scipy.linalg.tests.test_decomp.test_eigvals)
>FAIL: check_simple_tr 
>(scipy.linalg.tests.test_decomp.test_eigvals)
>FAIL: test_explicit (scipy.tests.test_odr.test_odr)
>FAIL: test_multi (scipy.tests.test_odr.test_odr)
>FAILED (failures=7)
> 
> Could you please have a look at my installation logs 
>(which are attached to
> this mail), and help me installing Scipy :
> 
>    - I'm not used to compile on that Linux64 box. I've 
>added options like
> '-fPIC and -m 64' ,
>      but i'm not sure this is enough. Have you got any 
>advices ?
> 
>    - scipy.linalg seems to have some pb but I didn't 
>have any pb when
> running numpy.test(),
>      What are the differences between scipy.linalg and 
>numpy.linalg ?
>      Why do I have pb with scipy and not with numpy ?
> 
>    - Besides, I did'nt have any pb compiling lapack and 
>atlas, but I see pb
> with lapack.
>      Any Idea on where it comes from ?
> 
>    - I see pb with odr, and I'm not sure to need this 
>module. If I don't
> succeed in solving this pb, is there a way to 
>desactivate this package, and
> prevent someone to use a "buggy" or "bad-installed" 
>module ?
> 
> Thanks
> 
> -- 
> LB

This is a known issue. See 
http://projects.scipy.org/scipy/scipy/ticket/375

Cheers,
  
              Nils
  



More information about the SciPy-User mailing list