Ian, Thanks for the tip. My env just had PY_VER set to 2.7, so I added PY_VERSION as well, but it didn't resolve the problem. I stepped through the code using pdb and get the following:
> /Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/engine/__init__.py(636)addRegion() -> engine.Network.addRegion(self, name, nodeType, nodeParams) (Pdb) s --Call-- > /Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/bindings/engine_internal.py(1009)addRegion() -> def addRegion(*args, **kwargs): (Pdb) n > /Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/bindings/engine_internal.py(1011)addRegion() -> return _engine_internal.Network_addRegion(*args, **kwargs) (Pdb) s ImportError: numpy.core.multiarray failed to import ERROR: Could not get valid spec for Region: py.RecordSensor [/Users/keith/Development/nupic/nta/engine/RegionImplFactory.cpp line 452] RuntimeError: RuntimeE...Sensor',) Segmentation fault Since it seg faults, it does seem to be a problem with how the numpy/core/multiarray.so lib was built, but I don't see any errors in the stdout.txt file from the build. Did the numpy version that is being used change recently? I'm still trying to figure out why this used to work before last week. Thanks, Matt On Sep 16, 2013, at 1:46 PM, Ian Danforth <[email protected]> wrote: > Matt, > > Austin found the cause of my issue, perhaps it is yours as well. For one > reason or another $PY_VERSION was not set in my environment. Once I 'export > PY_VERSION=2.7' things started working. Apparently numpy has switches all > over the place that rely on that env variable. > > Ian > > > On Mon, Sep 16, 2013 at 11:34 AM, Marek Otahal <[email protected]> wrote: > Thanks for reply Matt, > > that is, in deed, weird.. > Does python have some equivalent to set +x in shell? > Or could you try following the order by hand? and see where the err happens? > > > On Mon, Sep 16, 2013 at 8:04 PM, Matt Keith <[email protected]> wrote: > Mark, > > Yes, I have rebuilt all of the python modules via pip from the pip-caches > versions. What's weird is that numpy and multiarray seem to work fine when I > try to load and access them by hand: > >>> import numpy.core.multiarray > >>> numpy.core.multiarray.__version__ > '3.1' > > However, it fails when I try to run the nupic example. Is there a log > somewhere that might give more details about error? > > Thanks, > > Matt > > On Sep 16, 2013, at 11:36 AM, Marek Otahal <[email protected]> wrote: > >> Hi, >> there's no need for scipy imho. >> The hotgym example is being checked by Travis every pull, so the issue is >> reduced to OSX problems with any of the lib. Like Greg and Austin said, >> please verify with all python modules rebuilt (if possible, use the pip >> versions provided by nupic). Also you can test in virtualenv if that solves >> the problem for you.. >> >> >> On Mon, Sep 16, 2013 at 7:12 PM, Matt Keith <[email protected]> wrote: >> Greg, >> >> Thanks for the info. I do have pip installed already. Here is the version >> info: >> pip 1.4 from /Library/Python/2.7/site-packages/pip-1.4-py2.7.egg (python 2.7) >> >> Matt >> >> On Sep 16, 2013, at 9:37 AM, Tao Effect <[email protected]> wrote: >> >>> Installing Python through Homebrew will also install pip, which I think is >>> needed anyway to retrieve the dependencies. >>> >>> - Greg >>> >>> -- >>> Please do not email me anything that you are not comfortable also sharing >>> with the NSA. >>> >>> On Sep 16, 2013, at 11:34 AM, Matt Keith <[email protected]> wrote: >>> >>>> Ian, >>>> >>>> I do not have scipy installed on my system. Here is the output for me: >>>> >>>> Python 2.7.2 (default, Oct 11 2012, 20:14:37) >>>> [GCC 4.2.1 Compatible Apple Clang 4.0 (tags/Apple/clang-418.0.60)] on >>>> darwin >>>> Type "help", "copyright", "credits" or "license" for more information. >>>> >>> import numpy >>>> >>> import matplotlib >>>> >>> import scipy >>>> Traceback (most recent call last): >>>> File "<stdin>", line 1, in <module> >>>> ImportError: No module named scipy >>>> >>> numpy.__version__ >>>> '1.7.1' >>>> >>> matplotlib.__version__ >>>> '0.98.5.2' >>>> >>>> I had this working before and the only thing that changed on my system was >>>> an update of the nupic source, so it must have been a change in the build >>>> settings, process or requirements. I have done cleanbuild.sh and tried >>>> removing $NTA and $NTABUILD dirs by hand, but get the same error. >>>> >>>> I would rather not go the homebrew route. I don't think that should be a >>>> requirement for other contributors and don't want to maintain yet another >>>> development environment on my system. Hopefully, we can just find the >>>> issue and resolve it. >>>> >>>> Thanks, >>>> >>>> Matt >>>> >>>> On Sep 16, 2013, at 8:10 AM, Tao Effect <[email protected]> wrote: >>>> >>>>> Did you install Python through Homebrew already? >>>>> >>>>> -- >>>>> Please do not email me anything that you are not comfortable also sharing >>>>> with the NSA. >>>>> >>>>> On Sep 16, 2013, at 10:06 AM, Ian Danforth <[email protected]> >>>>> wrote: >>>>> >>>>>> I started by removing $NTA and $NUPIC and doing a fresh checkout and >>>>>> have done a clean build several times. Rather than go to the extreme >>>>>> length of reinstalling python, I'd like to track down the root cause of >>>>>> this issue so that a fix can be automated. >>>>>> >>>>>> Ian >>>>>> >>>>>> >>>>>> On Mon, Sep 16, 2013 at 5:43 AM, Tao Effect <[email protected]> >>>>>> wrote: >>>>>> These errors look identical to what I was running into and managed to >>>>>> fix in a recent pull request that was merged a few weeks ago. >>>>>> >>>>>> I would recommend installing Python using Homebrew and then doing a >>>>>> clean build of NuPIC (i.e. deleting the nta folder). >>>>>> >>>>>> Kind regards, >>>>>> Greg Slepak >>>>>> >>>>>> — @taoffect on GitHub and elsewhere >>>>>> >>>>>> -- >>>>>> Please do not email me anything that you are not comfortable also >>>>>> sharing with the NSA. >>>>>> >>>>>> On Sep 16, 2013, at 12:15 AM, Ian Danforth <[email protected]> >>>>>> wrote: >>>>>> >>>>>>> Matt, >>>>>>> >>>>>>> I ran 'sudo pip unintsall matplotlib' and this made python fall back >>>>>>> to the version supplied by NuPIC, which is version .98.5.2 (released >>>>>>> 2008!). And that let me import pylab and get run_tests.sh complete. >>>>>>> Going to rebuild and see if it clears up any other issues. >>>>>>> >>>>>>> Ian >>>>>>> >>>>>>> >>>>>>> On Sun, Sep 15, 2013 at 9:02 PM, Ian Danforth >>>>>>> <[email protected]> wrote: >>>>>>> Matt, >>>>>>> >>>>>>> Can you reply with your output for these versions if you have them? >>>>>>> Numpy, matplotlib, scipy, and scikit-learn are not happy on my system >>>>>>> and you might be in a similar situation. >>>>>>> >>>>>>> Ian >>>>>>> >>>>>>> >>> import numpy >>>>>>> >>> import matplotlib >>>>>>> >>> import scipy >>>>>>> >>> numpy.__version__ >>>>>>> '1.7.1' >>>>>>> >>> matplotlib.__version__ >>>>>>> '1.3.x' >>>>>>> >>> scipy.__version__ >>>>>>> '0.13.0.dev-c31f167' >>>>>>> >>> import sklearn >>>>>>> RuntimeError: module compiled against API version 8 but this version of >>>>>>> numpy is 7 >>>>>>> Traceback (most recent call last): >>>>>>> File "<stdin>", line 1, in <module> >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scikit_learn-0.14_git-py2.7-macosx-10.8-intel.egg/sklearn/__init__.py", >>>>>>> line 32, in <module> >>>>>>> from .base import clone >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scikit_learn-0.14_git-py2.7-macosx-10.8-intel.egg/sklearn/base.py", >>>>>>> line 10, in <module> >>>>>>> from scipy import sparse >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scipy-0.13.0.dev_c31f167_20130307-py2.7-macosx-10.8-intel.egg/scipy/sparse/__init__.py", >>>>>>> line 182, in <module> >>>>>>> from .csr import * >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scipy-0.13.0.dev_c31f167_20130307-py2.7-macosx-10.8-intel.egg/scipy/sparse/csr.py", >>>>>>> line 15, in <module> >>>>>>> from .sparsetools import csr_tocsc, csr_tobsr, csr_count_blocks, \ >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scipy-0.13.0.dev_c31f167_20130307-py2.7-macosx-10.8-intel.egg/scipy/sparse/sparsetools/__init__.py", >>>>>>> line 5, in <module> >>>>>>> from .csr import * >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scipy-0.13.0.dev_c31f167_20130307-py2.7-macosx-10.8-intel.egg/scipy/sparse/sparsetools/csr.py", >>>>>>> line 26, in <module> >>>>>>> _csr = swig_import_helper() >>>>>>> File >>>>>>> "/Library/Python/2.7/site-packages/scipy-0.13.0.dev_c31f167_20130307-py2.7-macosx-10.8-intel.egg/scipy/sparse/sparsetools/csr.py", >>>>>>> line 22, in swig_import_helper >>>>>>> _mod = imp.load_module('_csr', fp, pathname, description) >>>>>>> ImportError: numpy.core.multiarray failed to import >>>>>>> >>> >>>>>>> >>>>>>> >>>>>>> >>>>>>> On Sun, Sep 15, 2013 at 8:48 PM, Ian Danforth >>>>>>> <[email protected]> wrote: >>>>>>> Matt, >>>>>>> >>>>>>> I'm also having issues that end up in "RuntimeError: Could not get >>>>>>> valid spec for Region: py.RecordSensor" >>>>>>> >>>>>>> One thing I see is that "import pylab" fails for me with the same error >>>>>>> you're seeing, even though importing multiarray in the normal way >>>>>>> works. Looking into that. >>>>>>> >>>>>>> Ian >>>>>>> >>>>>>> >>>>>>> On Sun, Sep 15, 2013 at 8:31 PM, Matt Keith <[email protected]> wrote: >>>>>>> I used to have a working NuPIC setup on my rMBP running OS X 10.8.4 and >>>>>>> python 2.7, but last week I pulled the latest code from github and now >>>>>>> I get an import error when I run any client code. Everything builds >>>>>>> fine and it even passes htmtest and run_tests.sh. However, when I run >>>>>>> the hotgym example now I get the following error: >>>>>>> >>>>>>> -->python $NUPIC/examples/opf/clients/hotgym/hotgym.py >>>>>>> ImportError: numpy.core.multiarray failed to import >>>>>>> ERROR: Could not get valid spec for Region: py.RecordSensor >>>>>>> [/Users/keith/Development/nupic/nta/engine/RegionImplFactory.cpp line >>>>>>> 452] >>>>>>> Traceback (most recent call last): >>>>>>> File >>>>>>> "/Users/keith/Development/nupic/examples/opf/clients/hotgym/hotgym.py", >>>>>>> line 94, in <module> >>>>>>> runHotgym() >>>>>>> File >>>>>>> "/Users/keith/Development/nupic/examples/opf/clients/hotgym/hotgym.py", >>>>>>> line 65, in runHotgym >>>>>>> model = createModel() >>>>>>> File >>>>>>> "/Users/keith/Development/nupic/examples/opf/clients/hotgym/hotgym.py", >>>>>>> line 60, in createModel >>>>>>> return ModelFactory.create(model_params.MODEL_PARAMS) >>>>>>> File >>>>>>> "/Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/frameworks/opf/modelfactory.py", >>>>>>> line 76, in create >>>>>>> return modelClass(**modelConfig['modelParams']) >>>>>>> File >>>>>>> "/Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/frameworks/opf/clamodel.py", >>>>>>> line 203, in __init__ >>>>>>> anomalyParams) >>>>>>> File >>>>>>> "/Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/frameworks/opf/clamodel.py", >>>>>>> line 1078, in __createCLANetwork >>>>>>> n.addRegion("sensor", "py.RecordSensor", >>>>>>> json.dumps(dict(verbosity=sensorParams['verbosity']))) >>>>>>> File >>>>>>> "/Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/engine/__init__.py", >>>>>>> line 636, in addRegion >>>>>>> engine.Network.addRegion(self, name, nodeType, nodeParams) >>>>>>> File >>>>>>> "/Users/keith/Development/nta/eng/lib/python2.7/site-packages/nupic/bindings/engine_internal.py", >>>>>>> line 1011, in addRegion >>>>>>> return _engine_internal.Network_addRegion(*args, **kwargs) >>>>>>> RuntimeError: Could not get valid spec for Region: py.RecordSensor >>>>>>> >>>>>>> Is anyone else seeing this problem with the latest code? Any ideas as >>>>>>> to why the multiarray lib won't load? >>>>>>> >>>>>>> The .so file was built and exists in the $NAT/eng dir at >>>>>>> -rwxr-xr-x 1 keith staff 2921728 Sep 15 20:47 >>>>>>> lib/python2.7/site-packages/numpy/core/multiarray.so >>>>>>> And the module seems to load fine by hand: >>>>>>> >>> import numpy >>>>>>> >>> import numpy.core.multiarray >>>>>>> >>> numpy.__version__ >>>>>>> '1.7.1' >>>>>>> >>> numpy.core.multiarray.__version__ >>>>>>> '3.1' >>>>>>> >>>>>>> Let me know if you would like any output from the stdout.txt file from >>>>>>> the build. >>>>>>> >>>>>>> Thanks, >>>>>>> >>>>>>> Matt >>>>>>> _______________________________________________ >>>>>>> nupic mailing list >>>>>>> [email protected] >>>>>>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >>>>>>> >>>>>>> >>>>>>> >>>>>>> _______________________________________________ >>>>>>> nupic mailing list >>>>>>> [email protected] >>>>>>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> nupic mailing list >>>>>> [email protected] >>>>>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >>>>>> >>>>>> >>>>>> _______________________________________________ >>>>>> nupic mailing list >>>>>> [email protected] >>>>>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >>>>> >>>>> _______________________________________________ >>>>> nupic mailing list >>>>> [email protected] >>>>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >>>> >>>> _______________________________________________ >>>> nupic mailing list >>>> [email protected] >>>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >>> >>> _______________________________________________ >>> nupic mailing list >>> [email protected] >>> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >> >> >> _______________________________________________ >> nupic mailing list >> [email protected] >> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org >> >> >> >> >> -- >> Marek Otahal :o) >> _______________________________________________ >> nupic mailing list >> [email protected] >> http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org > > > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org > > > > > -- > Marek Otahal :o) > > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org > > > _______________________________________________ > nupic mailing list > [email protected] > http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
_______________________________________________ nupic mailing list [email protected] http://lists.numenta.org/mailman/listinfo/nupic_lists.numenta.org
