Jupyter（IPython）笔记本numpy / pandas / matplotlib错误（FreeBSD）

Question

I am trying to set up a Jupyter notebook server at home. It has taken me a long time, but I have build and installed Python 3.4 and all the required packages from FreeBSD ports successfully. The notebook server is up and running fine, except every time when I try to import numpy:

In[1]: import numpy

The following errors occur:

ImportError                               Traceback (most recent call last)
<ipython-input-1-5a0bd626bb1d> in <module>()
----> 1 import numpy

/usr/local/lib/python3.4/site-packages/numpy/__init__.py in <module>()
    178         return loader(*packages, **options)
    179 
--> 180     from . import add_newdocs
    181     __all__ = ['add_newdocs',
    182                'ModuleDeprecationWarning',

/usr/local/lib/python3.4/site-packages/numpy/add_newdocs.py in <module>()
     11 from __future__ import division, absolute_import, print_function
     12 
---> 13 from numpy.lib import add_newdoc
     14 
     15 ###############################################################################

/usr/local/lib/python3.4/site-packages/numpy/lib/__init__.py in <module>()
      6 from numpy.version import version as __version__
      7 
----> 8 from .type_check import *
      9 from .index_tricks import *
     10 from .function_base import *

/usr/local/lib/python3.4/site-packages/numpy/lib/type_check.py in <module>()
      9            'common_type']
     10 
---> 11 import numpy.core.numeric as _nx
     12 from numpy.core.numeric import asarray, asanyarray, array, isnan, \
     13                 obj2sctype, zeros

/usr/local/lib/python3.4/site-packages/numpy/core/__init__.py in <module>()
     12         os.environ[envkey] = '1'
     13         env_added.append(envkey)
---> 14 from . import multiarray
     15 for envkey in env_added:
     16     del os.environ[envkey]

ImportError: /lib/libgcc_s.so.1: version GCC_4.6.0 required by /usr/local/lib/gcc48/libgfortran.so.3 not found

The error messages for importing pandas and matplotlib are different, but I suspect that has something to do with this numpy import error.

Strangely, all 3 packages work fine in Python and IPython consoles with no problems at all! I have googled and made the following attempts:

    delete and reinstall numpy -> no change
    append numpy directory to sys.path -> no change
    install a lot of other external packages just to see if it's only related to numpy -> they are all working fine in both consoles and notebook, except scipy giving some error related to numpy

Thank you for your help!

Answer 1

My gcc is version 4.2.1. I have fixed this by setting the LD_LBRARY_PATH to /usr/local/lib/gcc48. gcc48 is already installed in my system. To avoid setting the path every time, I've added the following line to /.cshrc:

setenv  LD_LIBRARY_PATH /usr/local/lib/gcc48

edit: This won't work is you want to start the notebook server automatically by adding to crontab:

@reboot /usr/local/bin/jupyter-notebook

the same error appears when trying to import numpy and modules depending on numpy

I fixed this by making a copy of /usr/local/bin/jupyter-notebook and added the following lines:

    import sys
    import re
    ----------------- add these 2 lines below --------------
    import os
    os.environ['LD_LIBRARY_PATH'] = '/usr/local/lib/gcc48'
    ....

Add the new file to crontab instead of jupyter-notebook .

Answer 2

The issue is not with your python modules. The error message at the bottom, where it says ImportError: /lib/libgcc_s.so.1: version GCC_4.6.0 required by /usr/local/lib/gcc48/libgfortran.so.3 not found indicates that it's a dependency error with the Fortran library. Apparently it wants gcc 4.6 or higher, and apparently you have a lower version installed. Not being familiar with Python libraries or your setup, my guess is that it could be an issue with /usr/ports/devel/py-fortran . I would recommend checking the gcc version on your machine with gcc -v and whatever fortran-related ports you have installed with pkg info and then take it from there.