[Numpy-tickets] [NumPy] #370: oldnumeric.Int32 mapping on 32-bit platforms seems wrong

NumPy numpy-tickets at scipy.net
Thu Nov 2 04:48:12 CST 2006


#370: oldnumeric.Int32 mapping on 32-bit platforms seems wrong
-----------------------+----------------------------------------------------
 Reporter:  faltet     |       Owner:  somebody
     Type:  defect     |      Status:  new     
 Priority:  normal     |   Milestone:          
Component:  numpy.lib  |     Version:  devel   
 Severity:  normal     |    Keywords:          
-----------------------+----------------------------------------------------
 Hi,

 I think that, for 32-bit platforms, oldnumeric.Int32 should be mapped to a
 C long, instead as a C int as it is now. This is to be consistent with:

 {{{
 In [16]: numpy.zeros((1,), numpy.int32).dtype.num
 Out[16]: 7
 In [17]: numpy.zeros((1,), numpy.sctypeNA['Int32']).dtype.num
 Out[17]: 7
 In [18]: numpy.zeros((1,), oldnumeric.int32).dtype.num
 Out[18]: 7
 # Fine until here, but:
 In [19]: numpy.zeros((1,), oldnumeric.Int32).dtype.num
 Out[19]: 5  # !!
 }}}

 This can effectively be a source of problems with packages that use
 oldnumeric. scipy.ndimage is one of them that gets affected by this. See:

 http://aspn.activestate.com/ASPN/Mail/Message/scipy-user/3321095

 Regards,

-- 
Ticket URL: <http://projects.scipy.org/scipy/numpy/ticket/370>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.


More information about the Numpy-tickets mailing list