[SciPy-user] [Numpy-discussion] scipy 0.7.1 rc3

David Cournapeau david@ar.media.kyoto-u.ac...
Sat Jun 13 06:03:44 CDT 2009


josef.pktd@gmail.com wrote:
> On Fri, Jun 12, 2009 at 7:09 PM, <josef.pktd@gmail.com> wrote:
>   
>> On Fri, Jun 12, 2009 at 6:26 PM, George Nurser<gnurser@googlemail.com> wrote:
>>     
>>> Hi,
>>> I have just two failures on svn r 5835, which I believe is the same as
>>> 0.7.1.rc3.
>>>       
>> Are you sure you are using the tag version and not the trunk version?
>> I think the test failure with ksone was created when I corrected and
>> tightened the test in revision 5831 in the trunk,
>>
>> the precision in 0.7.1 is set low enough that it should not report
>> this as a failure.
>>
>> I didn't catch the error in the changes to the trunk because I skipped
>> the slow tests. This also means there is a bug or imprecision in ksone
>> or special.smirnov
>>
>> Josef
>>
>>
>>     
>>> OS x 10.5.7,python.org 2.5.2, numpy svn 7044, gfortran 4.3.2, apple gcc 4.0.1.
>>> scipy built with
>>> python setup.py build_src build_clib config_fc --fcompiler=gnu95
>>> --f77flags=" -O3 -march=core2" build_ext config_fc --fcompiler=gnu95
>>> --f77flags=" -O3 -march=core2" build > & inst.log &
>>>
>>> scipy.test('10')
>>> [snip]
>>> =====================================================================
>>> FAIL: test_random_real (test_basic.TestSingleIFFT)
>>> ----------------------------------------------------------------------
>>> Traceback (most recent call last):
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/fftpack/tests/test_basic.py",
>>> line 205, in test_random_real
>>>    assert_array_almost_equal (y1, x)
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy/testing/utils.py",
>>> line 537, in assert_array_almost_equal
>>>    header='Arrays are not almost equal')
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy/testing/utils.py",
>>> line 395, in assert_array_compare
>>>    raise AssertionError(msg)
>>> AssertionError:
>>> Arrays are not almost equal
>>>
>>> (mismatch 0.900900900901%)
>>>  x: array([ 0.56936401 -1.61093638e-09j,  0.42778808 -9.66561764e-09j,
>>>        0.52186930 -1.07395759e-09j,  0.09524378 -1.19960097e-12j,
>>>        0.37034556 -1.55349547e-08j,  0.87382299 +7.88580223e-09j,...
>>>  y: array([ 0.56936431,  0.42778817,  0.52186954,  0.09524398,  0.37034538,
>>>        0.87382281,  0.11707421,  0.78889829,  0.92179215,  0.82394433,
>>>        0.23986711,  0.61445147,  0.64488971,  0.77991045,  0.91312933,...
>>>
>>> ======================================================================
>>> FAIL: test_continuous_basic.test_cont_basic_slow(<scipy.stats.distributions.ksone_gen
>>> object at 0x15d6f2d0>, (22,), 'ksone')
>>> ----------------------------------------------------------------------
>>> Traceback (most recent call last):
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/nose-0.10.3-py2.5.egg/nose/case.py",
>>> line 182, in runTest
>>>    self.test(*self.arg)
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/scipy/stats/tests/test_continuous_basic.py",
>>> line 290, in check_cdf_ppf
>>>    ' - cdf-ppf roundtrip')
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy/testing/utils.py",
>>> line 262, in assert_almost_equal
>>>    return assert_array_almost_equal(actual, desired, decimal, err_msg)
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy/testing/utils.py",
>>> line 537, in assert_array_almost_equal
>>>    header='Arrays are not almost equal')
>>>  File "/Library/Frameworks/Python.framework/Versions/2.5/lib/python2.5/site-packages/numpy/testing/utils.py",
>>> line 395, in assert_array_compare
>>>    raise AssertionError(msg)
>>> AssertionError:
>>> Arrays are not almost equal
>>> ksone - cdf-ppf roundtrip
>>> (mismatch 33.3333333333%)
>>>  x: array([ 0.   ,  0.5  ,  0.999])
>>>  y: array([ 0.001,  0.5  ,  0.999])
>>>
>>> ----------------------------------------------------------------------
>>> Ran 4273 tests in 588.327s
>>>
>>> --George.
>>>
>>>
>>>
>>> 2009/6/12 David Cournapeau <david@ar.media.kyoto-u.ac.jp>:
>>>       
>>>> Hi,
>>>>
>>>>    I have uploaded the binaries and source tarballs for 0.7.1rc3. The
>>>> rc3 fixes some issues in scipy.special, which caused wrong
>>>> behavior/crashes on some platforms. Hopefully, this will be the 0.7.1
>>>> release,
>>>>
>>>> cheers,
>>>>
>>>> David
>>>>
>>>> =========================
>>>> SciPy 0.7.1 Release Notes
>>>> =========================
>>>>
>>>> .. contents::
>>>>
>>>> SciPy 0.7.1 is a bug-fix release with no new features compared to 0.7.0.
>>>>
>>>> scipy.io
>>>> ========
>>>>
>>>> Bugs fixed:
>>>>
>>>> - Several fixes in Matlab file IO
>>>>
>>>> scipy.odr
>>>> =========
>>>>
>>>> Bugs fixed:
>>>>
>>>> - Work around a failure with Python 2.6
>>>>
>>>> scipy.signal
>>>> ============
>>>>
>>>> Memory leak in lfilter have been fixed, as well as support for array object
>>>>
>>>> Bugs fixed:
>>>>
>>>> - #880, #925: lfilter fixes
>>>> - #871: bicgstab fails on Win32
>>>>
>>>>
>>>> scipy.sparse
>>>> ============
>>>>
>>>> Bugs fixed:
>>>>
>>>> - #883: scipy.io.mmread with scipy.sparse.lil_matrix broken
>>>> - lil_matrix and csc_matrix reject now unexpected sequences,
>>>>  cf. http://thread.gmane.org/gmane.comp.python.scientific.user/19996
>>>>
>>>> scipy.special
>>>> =============
>>>>
>>>> Several bugs of varying severity were fixed in the special functions:
>>>>
>>>> - #503, #640: iv: problems at large arguments fixed by new implementation
>>>> - #623: jv: fix errors at large arguments
>>>> - #679: struve: fix wrong output for v < 0
>>>> - #803: pbdv produces invalid output
>>>> - #804: lqmn: fix crashes on some input
>>>> - #823: betainc: fix documentation
>>>> - #834: exp1 strange behavior near negative integer values
>>>> - #852: jn_zeros: more accurate results for large s, also in
>>>> jnp/yn/ynp_zeros
>>>> - #853: jv, yv, iv: invalid results for non-integer v < 0, complex x
>>>> - #854: jv, yv, iv, kv: return nan more consistently when out-of-domain
>>>> - #927: ellipj: fix segfault on Windows
>>>> - #946: ellpj: fix segfault on Mac OS X/python 2.6 combination.
>>>> - ive, jve, yve, kv, kve: with real-valued input, return nan for
>>>> out-of-domain
>>>>  instead of returning only the real part of the result.
>>>>
>>>> Also, when ``scipy.special.errprint(1)`` has been enabled, warning
>>>> messages are now issued as Python warnings instead of printing them to
>>>> stderr.
>>>>
>>>>
>>>> scipy.stats
>>>> ===========
>>>>
>>>> - linregress, mannwhitneyu, describe: errors fixed
>>>> - kstwobign, norm, expon, exponweib, exponpow, frechet, genexpon, rdist,
>>>>  truncexpon, planck: improvements to numerical accuracy in distributions
>>>>
>>>> Windows binaries for python 2.6
>>>> ===============================
>>>>
>>>> python 2.6 binaries for windows are now included. The binary for python 2.5
>>>> requires numpy 1.2.0 or above, and and the one for python 2.6 requires numpy
>>>> 1.3.0 or above.
>>>>
>>>> Universal build for scipy
>>>> =========================
>>>>
>>>> Mac OS X binary installer is now a proper universal build, and does not
>>>> depend
>>>> on gfortran anymore (libgfortran is statically linked). The python 2.5
>>>> version
>>>> of scipy requires numpy 1.2.0 or above, the python 2.6 version requires
>>>> numpy
>>>> 1.3.0 or above.
>>>>
>>>> Checksums
>>>> =========
>>>>
>>>> 9dd5af43cc26ae6d38a13b373ba430fa
>>>> release/installers/scipy-0.7.1rc3-py2.6-python.org.dmg
>>>> 290c2e056fda1f86dfa9f3a76d207a8c
>>>> release/installers/scipy-0.7.1rc3-win32-superpack-python2.6.exe
>>>> d582dff7535d2b64a097fb4bfbc75d09
>>>> release/installers/scipy-0.7.1rc3-win32-superpack-python2.5.exe
>>>> a19400ccfd65d1a0a5030848af6f78ea  release/installers/scipy-0.7.1rc3.tar.gz
>>>> d4ebf322c62b09c4ebaad7b67f92d032  release/installers/scipy-0.7.1rc3.zip
>>>> a0ea0366b178a7827f10a480f97c3c47
>>>> release/installers/scipy-0.7.1rc3-py2.5-python.org.dmg
>>>>
>>>>         
>
> Why are there now 310 known fail, it was only a few before, what I remember?
> now errors or failures.
>
> Josef
>
> Ran 4168 tests in 383.109s
> OK (KNOWNFAIL=310, SKIP=29)
>
> Winxp, sse2, 32bit
>   

That may be due to the 0.11 version of nose. I have noticed for quite
some time some significant discrepancy between windows and other
platforms in the number of found tests, and this always bothered me.
There have not been more known failures on 0.7.1 compared to 0.7.0.

cheers,

David


More information about the SciPy-user mailing list