[SciPy-dev] Building on Woody(Debian): good news, bad news.

eric eric at scipy.org
Sat Mar 23 00:47:08 EST 2002


Hey Prabhu,

[snip stuff about fixing Debian Python]

Thanks Prabhu for riding herd on this one.

>
> After I built scipy the tests seem to run but I get the following
> error:
>
> In [1]: import scipy
> /usr/local/lib/python2.1/site-packages/scipy/integrate/vode.so: undefined
symbol: f2py_report_on_exit

I just got this when building on RH 7.1 today also.  You can get rid of it by
re-compiling with F2PY_REPORT_ATEXIT_DISABLE defined.

I got rid of it by edit my f2py2e/src/fortranobject.h header to have disable as
the default.

Pearu, this seems to cause problems on a lot of platforms.  Any chance you can
change the reporting stuff to default to off in f2py?

>
> I dont know why this happens.  When I run the tests I get 4 failures.
> But it does run and does not segfault anymore. :)
>
> In [2]: scipy.test()
> [snip]
> ======================================================================
> FAIL: check_default_cols (test_misc.test_mean)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/usr/local/lib/python2.1/site-packages/scipy/tests/test_misc.py", line
53, in check_default_cols
>     assert_array_equal(val,desired)
>   File "/usr/local/lib/python2.1/site-packages/scipy_test/scipy_test.py", line
301, in assert_array_equal
>     assert alltrue(ravel(reduced)),\
> AssertionError:
> Arrays are not equal:
>
> ======================================================================
> FAIL: check_qnan (test_misc.test_isnan)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/usr/local/lib/python2.1/site-packages/scipy/tests/test_misc.py", line
112, in check_qnan
>     assert(isnan(log(-1.)) == 1)
> AssertionError
> ======================================================================
>
> ======================================================================
> FAIL: check_qnan (test_misc.test_isfinite)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/usr/local/lib/python2.1/site-packages/scipy/tests/test_misc.py", line
132, in check_qnan
>     assert(isfinite(log(-1.)) == 0)
> AssertionError
> ======================================================================
> FAIL: check_qnan (test_misc.test_isinf)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>   File "/usr/local/lib/python2.1/site-packages/scipy/tests/test_misc.py", line
157, in check_qnan
>     assert(isnan(log(-1.)) == 1)
> AssertionError
> ----------------------------------------------------------------------
> Ran 262 tests in 3.281s
>
> FAILED (failures=4)
>

I get the exact same errors on RH7.1.  The NaN stuff is still flaky.  We get it
working on one platform and it breaks on another.  Not sure about the first
failure -- it looks like that one should pass.

eric






More information about the SciPy-Dev mailing list