Closed ghost closed 10 years ago
its py3.3 broken with 3.1 which is prob coming out soon; I guess I can just put up a warning in the docs/install. The only problem is that its going to become the default soon.I could doc it I guess.
@cgohlke can you incorporate that commit in the 3.3 build? (I believe this is ONLY a windows py3.3. issue w/ pytables 3.1)
Hi guys, building 0.13.1 now and still getting the old
======================================================================
ERROR: test_select_with_many_inputs (pandas.io.tests.test_pytables.TestHDFStore)
----------------------------------------------------------------------
Traceback (most recent call last):
File "/tmp/buildd/pandas-0.13.0+git464-g15a8ff7/debian/tmp/usr/lib/python2.7/dist-packages/pandas/io/tests/test_pytables.py", line 2999, in test_select_with_many_inputs
result = store.select('df', [Term("ts>=Timestamp('2012-02-01')"),Term('users=selector')])
File "/tmp/buildd/pandas-0.13.0+git464-g15a8ff7/debian/tmp/usr/lib/python2.7/dist-packages/pandas/io/pytables.py", line 643, in select
raise KeyError('No object named %s in the file' % key)
KeyError: 'No object named df in the file'
that is on elderly ubuntu with
INSTALLED VERSIONS
------------------
commit: None
python: 2.7.3.final.0
python-bits: 64
OS: Linux
OS-release: 3.2.0-4-amd64
machine: x86_64
processor: x86_64
byteorder: little
LC_ALL: C
LANG: C
pandas: 0.13.0.dev
Cython: 0.15.1
numpy: 1.6.1
scipy: 0.9.0
statsmodels: 0.5.0
IPython: None
sphinx: 1.1.3
patsy: 0.2.1
scikits.timeseries: None
dateutil: 1.5
pytz: 2012c
bottleneck: None
tables: 2.3.1
numexpr: 1.4.2
matplotlib: 1.1.1rc
openpyxl: 1.7.0
xlrd: 0.6.1
xlwt: 0.7.2
xlsxwriter: None
sqlalchemy: None
lxml: None
bs4: 4.0.2
html5lib: 0.90
bq: None
apiclient: None
please advise
@yarikoptic numexpr is too old for pytables to deal; THEY don't even support < 2.0 I think.
will have to disable this for pytables stuff. Just skip it.
btw -- fun part that on Debian wheezy it fails only for 64bit but not 32:
neurodebian@head2:~/deb/builds/pandas/0.13.1-1$ grep test_select_with_many_inputs pandas_0.13.1-1~nd70+1_*.build
pandas_0.13.1-1~nd70+1_amd64.build:test_select_with_many_inputs (pandas.io.tests.test_pytables.TestHDFStore) ... ERROR
pandas_0.13.1-1~nd70+1_amd64.build:ERROR: test_select_with_many_inputs (pandas.io.tests.test_pytables.TestHDFStore)
pandas_0.13.1-1~nd70+1_amd64.build: File "/tmp/buildd/pandas-0.13.1/debian/tmp/usr/lib/python2.7/dist-packages/pandas/io/tests/test_pytables.py", line 3029, in test_select_with_many_inputs
pandas_0.13.1-1~nd70+1_i386.build:test_select_with_many_inputs (pandas.io.tests.test_pytables.TestHDFStore) ... ok
tables 2.3.1-3, numexpr 2.0.1-3
can u give me a login to that machine? the chroot u have me for sparc is only 64/bit right?
yes -- this error happens on intels... unfortunately access to that particular machine would be problematic. But now (in the evening) I will try to follow my own http://neuro.debian.net/blog/2012/2012-04-14_ndtools.html within NeuroDebian VM to provide you with the ultimate 'replicate-yarikoptic' environment/instructions, ok?
actually in that case it is just amd64 wheezy, so should appear in stock NeuroDebian wheezy VM... will test that first (running to pick up kids now)
I cannot repro this: numexpr 1.4.2 / 2.0.1, tables 2.3.1, numpy 1.6.1, cython 0.15.1
Iirc Cython was 0.20 in my case... yet to get to the laptop
jreback notifications@github.com wrote:
I cannot repro this: numexpr 1.4.2 / 2.0.1, tables 2.3.1, numpy 1.6.1, cython 0.15.1
Reply to this email directly or view it on GitHub: https://github.com/pydata/pandas/issues/6103#issuecomment-34009306
Sent from a phone which beats iPhone.
it might be more fun -- replicated while building package on NeuroDebian VM, but then it didn't fail when I just ran it with nosetests directly, so might be particularities of environment or smth like that -- will check and report back
@yarikoptic , can I get ssh access to the sparc box? just need to gather some system info.
@y-p: sparc -- sure. send me your desired login name and public portion of ssh key with a passphrase ;-) this issue smells like a race condition somewhere. Seems to come up only when running full set of tests (nosetests -s -v pandas), and not 100% always, but never if I just run that test separately. I do not have enough brain bandwidth (nor clue where to dig) so I guess I will just catch exception at that point and skip if that particular one comes up
Will email you my ssh public key. My passphrase is "1-2-3-4", but don't tell anyone.
@cgohlke can you update the windows 3.3 build with the commit https://github.com/pydata/pandas/commit/9bc7bac27242a7192a1194a5eac943c821c0dd68
@cgohlke thanks....I meant could you actually update the build on your site? (just for 3.3, 64/32 bit)
I already updated the pandas installers on my website on February 3rd.
@cgohlke ok..great thanks!
Just FYI, 0.13.1 still isn't showing up on PyPI.
shortly
pypi is now updated and ANN is sent!
We're due to release 0.13.1 in a few days (or as soon as we clear the queue), as a bugfix release stabilising 0.13.0, allowing us to start on the 3 month release cycle for 0.14.
Anything that isn't in 0.13.1 will have to wait until then to hit pypi.
Any objections, warnings or notes?
cc @jseabold , @yarikoptic, @juliantaylor, @takluyver