[pymvpa] issue with pymvpa tutorial on neurodebian VM

[Emanuele Olivetti]

On 04/05/2012 12:46 PM, Emanuele Olivetti wrote:
> Hi,
>
> this morning I attempted the first command of the PyMVPA tutorial
> on an updated NeuroDebian virtual machine and observed this:
> -------
> brain at neurodebian:~$ python
> Python 2.6.6 (r266:84292, Dec 26 2010, 22:31:48)
> [GCC 4.4.5] on linux2
> Type "help", "copyright", "credits" or "license" for more information.
> >>> from mvpa2.tutorial_suite import *
> /usr/lib/pymodules/python2.6/scikits/learn/__init__.py:2: UserWarning: scikits.learn 
> namespace is deprecated, please use sklearn instead
>   warnings.warn('scikits.learn namespace is deprecated, please use sklearn instead')
> /usr/lib/pymodules/python2.6/sklearn/hmm.py:24: UserWarning: sklearn.hmm is orphaned, 
> undocumented and has known numerical stability issues. If nobody volunteers to write 
> documentation and make it more stable, this module will be removed in version 0.11.
>   warnings.warn('sklearn.hmm is orphaned, undocumented and has known numerical'
> /usr/lib/pymodules/python2.6/scikits/learn/lda.py:2: UserWarning: scikits.learn 
> namespace is deprecated, please use sklearn instead
>   warnings.warn('scikits.learn namespace is deprecated, please use sklearn instead')
> /usr/lib/pymodules/python2.6/scikits/learn/grid_search.py:2: UserWarning: scikits.learn 
> namespace is deprecated, please use sklearn instead
>   warnings.warn('scikits.learn namespace is deprecated, please use sklearn instead')
> /usr/lib/pymodules/python2.6/scikits/learn/naive_bayes.py:2: UserWarning: scikits.learn 
> namespace is deprecated, please use sklearn instead
>   warnings.warn('scikits.learn namespace is deprecated, please use sklearn instead')
> /usr/lib/pymodules/python2.6/scikits/learn/neighbors.py:2: UserWarning: scikits.learn 
> namespace is deprecated, please use sklearn instead
>   warnings.warn('scikits.learn namespace is deprecated, please use sklearn instead')
> /usr/lib/pymodules/python2.6/scikits/learn/qda.py:2: UserWarning: scikits.learn 
> namespace is deprecated, please use sklearn instead
>   warnings.warn('scikits.learn namespace is deprecated, please use sklearn instead')
> Traceback (most recent call last):
>   File "<stdin>", line 1, in <module>
>   File "/usr/lib/pymodules/python2.6/mvpa2/tutorial_suite.py", line 15, in <module>
>     from mvpa2.suite import *
>   File "/usr/lib/pymodules/python2.6/mvpa2/suite.py", line 154, in <module>
>     if externals.exists('mdp ge 2.4'):
>   File "/usr/lib/pymodules/python2.6/mvpa2/base/externals.py", line 595, in exists
>     exec _KNOWN[dep]
>   File "<string>", line 1, in <module>
>   File "/usr/lib/pymodules/python2.6/mdp/__init__.py", line 166, in <module>
>     import nodes
>   File "/usr/lib/pymodules/python2.6/mdp/nodes/__init__.py", line 54, in <module>
>     del convolution_nodes
> NameError: name 'convolution_nodes' is not defined
> >>>
> -----
> As far as I understand, besides the warning on the scikits->sklearn renaming, there is
> an issue with MDP. I am digging it but in case you already know that I am
> sending this message immediately.
>
>

After a little digging I found out that the package python-mdp has some other
bigger issues: even "import mdp" fails. This is quite unexpected...

This tiny change in MDP seems to solve the issue of my previous message though.

-------------------------------------------
--- __init__.py.original    2012-04-05 07:46:13.000000000 -0400
+++ __init__.py    2012-04-05 07:48:40.000000000 -0400
@@ -51,17 +51,17 @@
  if numx_description == 'scipy':
      from convolution_nodes import Convolution2DNode
      __all__ += ['Convolution2DNode']
-    del convolution_nodes
+    # del convolution_nodes

  if config.has_shogun:
      from shogun_svm_classifier import ShogunSVMClassifier
      __all__ += ['ShogunSVMClassifier']
-    del shogun_svm_classifier
+    # del shogun_svm_classifier

  if config.has_libsvm:
      from libsvm_classifier import LibSVMClassifier
      __all__ += ['LibSVMClassifier']
-    del libsvm_classifier
+    # del libsvm_classifier

  if config.has_scikits:
      import scikits_nodes
--------------------------------------------

Best,

Emanuele