[Numpy-tickets] [NumPy] #219: NameError: PyArray_ContiguousFromObject on FC3; affects mtrand
NumPy
numpy-tickets at scipy.net
Tue Aug 1 12:39:35 CDT 2006
#219: NameError: PyArray_ContiguousFromObject on FC3; affects mtrand
-----------------------+----------------------------------------------------
Reporter: TomLoredo | Owner: somebody
Type: defect | Status: new
Priority: normal | Milestone:
Component: Other | Version:
Severity: normal | Keywords:
-----------------------+----------------------------------------------------
C code that calls PyArray_ContiguousFromObject (both my own and within
numpy)
compiles fine but gives a NameError at runtime on Fedora Core 3 (Python
2.4.3).
Calling random.set_state displays the bug. If I edit mtrand.pyx to change
its 3 PyArray_ContiguousFromObject calls to PyArray_ContiguousFromAny, run
generate_mtrand_c.py, and reinstall, the error goes away. A sample
IPython
session is copied below.
I deleted my previous numpy package from site-packages before installing
this one.
I do not have this problem on OS X (10.3.9, MacPy 2.4.1).
I am unclear of the status of PyArray_ContiguousFromObject. It is no
longer
mentioned in the Guide To Numpy. Is it deprecated?
-Tom Loredo
{{{
In [1]: import numpy
In [2]: numpy.__version__
Out[2]: '1.0b2.dev2941'
In [3]: from numpy import random
In [4]: s=random.get_state()
In [5]: s
Out[5]:
('MT19937',
array([-1367602364, 599621924, 1947610538, -686169956, -1451278673,
1648572372, -1175619180, -205273143, 13145610, 1292301198,
[snip]
-716751809, 622130452, 123887232, -2052494269]),
624)
In [6]: random.set_state(s)
---------------------------------------------------------------------------
exceptions.NameError Traceback (most
recent call last)
/home/loredo/research/exoplanets/rv/<ipython console>
/home/loredo/research/exoplanets/rv/mtrand.pyx in
mtrand.RandomState.set_state()
NameError: PyArray_ContiguousFromObject
}}}
--
Ticket URL: <http://projects.scipy.org/scipy/numpy/ticket/219>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.
More information about the Numpy-tickets
mailing list