[NumPy-Tickets] [NumPy] #1886: ndarray.data not compatible between python 2.7 and 3.2 versions

NumPy Trac numpy-tickets@scipy....
Wed Jun 29 08:27:52 CDT 2011


#1886: ndarray.data not compatible between python 2.7 and 3.2 versions
-------------------------------------------+--------------------------------
 Reporter:  axh                            |       Owner:  somebody   
     Type:  defect                         |      Status:  new        
 Priority:  normal                         |   Milestone:  Unscheduled
Component:  numpy.core                     |     Version:  1.5.1      
 Keywords:  compatibility, design clarity  |  
-------------------------------------------+--------------------------------
 I am running python 3.2 / 2.7.2 with the standards from Fedora 15

 numpy.x86_64
 1:1.5.1-1.fc15

 python3-numpy-f2py.x86_64
 1:1.5.1-1.fc15

 Here is the issue:

 with

 i4 = np.ndarray([1],dtype=np.int32)

 in 2,7 I have

 type(i4.data)
 <type 'buffer'>

 len(i4.data[:])
 4

 so the buffer is byte-size

 in 3.2 the data buffer comes in chunks of the size of the basic type

 len(i4.data[0])
 4

 len(i4.data)
 1

 type(i4.data)
 <class 'memoryview'>

 It would be most desirable to switch this back to a mode that provides
 byte-size buffer in the same way in both versions.  Especially, it would
 be good if the buffer type memory layout does not depend on the type of
 objects it stores.  It should be just a data buffer.

 -Alexander

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/1886>
NumPy <http://projects.scipy.org/numpy>
My example project


More information about the NumPy-Tickets mailing list