undefined symbol: gotoblas issue

[oztc]

my environments:

1. python
   Python 2.7.12+ (default, Nov 22 2016, 00:48:54)
   [GCC 6.2.1 20161119] on linux2
   Type "help", "copyright", "credits" or "license" for more information.

2. os

uname -a
Linux debian 3.16.0-4-amd64 #1 SMP Debian 3.16.7-ckt25-2 (2016-04-08) x86_64 GNU/Linux

3. the following log:

import numpy as np
File "/usr/lib/python2.7/dist-packages/numpy/init.py", line 153, in
from . import add_newdocs
File "/usr/lib/python2.7/dist-packages/numpy/add_newdocs.py", line 13, in
from numpy.lib import add_newdoc
File "/usr/lib/python2.7/dist-packages/numpy/lib/init.py", line 18, in
from .polynomial import *
File "/usr/lib/python2.7/dist-packages/numpy/lib/polynomial.py", line 19, in
from numpy.linalg import eigvals, lstsq, inv
File "/usr/lib/python2.7/dist-packages/numpy/linalg/init.py", line 50, in
from .linalg import *
File "/usr/lib/python2.7/dist-packages/numpy/linalg/linalg.py", line 29, in
from numpy.linalg import lapack_lite, _umath_linalg
ImportError: /usr/lib/libblas.so.3: undefined symbol: gotoblas

[brada4]

can you post 'pip freeze' and 'help (modules)' (attach .txt if more than dozen lines)
How did you compile/install OpenBLAS and numpy?
What does 'update-alternatives --list libblas.so.3' say?

[oztc]

1. pip freeze
   Traceback (most recent call last):
   File "/usr/local/bin/pip", line 5, in
   from pkg_resources import load_entry_point
   ImportError: No module named pkg_resources

2. update-alternatives --list libblas.so.3
   /usr/lib/libblas/libblas.so.3
   /usr/lib/openblas-base/libblas.so.3

[martin-frbg]

Try doing "sudo update-alternatives --set libblas.so.3 /usr/lib/openblas-base/libblas.so.3" to switch from the (unoptimized netlib reference ?) blas version currently installed as default on your system to the OpenBLAS that your numpy build expects.

[oztc]

ok, it is not a openblas bug, when i use anaconda python, now it can work.

thank you for your support.

[martin-frbg]

Then the anaconda version was probably built against the non-optimized blas - you can read about the options in https://github.com/scipy/scipy.org/blob/master/www/scipylib/building/linux.rst#debian--ubuntu
(I do hope that most if not all of the bugs that made them advise against using OpenBLAS two years ago
have been addressed in the meantime - in my experience, OpenBLAS is substantially faster than Atlas and the reference implementation with its lack of hardware-specific optimization is naturally the slowest.)

[oztc]

your advise and your link are very helpful, I will try it.

[brada4]

Can you close the issue?
You have ubuntu python in the text you posted. Ubuntu numpy (installed via APT) will respect update-alternatives.
Continuum anaconda (not in posted text) calls openblas 'nomkl' as opposed to default 'mkl', it is described in their numpy documents how to switch back and forth.