[NumPy-Tickets] [NumPy] #2037: genfromtxt with dtype different behavior in 32bit or 64bit numpy

[NumPy Trac]

#2037: genfromtxt with dtype different behavior in 32bit or 64bit numpy
------------------------+---------------------------------------------------
 Reporter:  risotto     |       Owner:  somebody   
     Type:  defect      |      Status:  new        
 Priority:  normal      |   Milestone:  Unscheduled
Component:  Other       |     Version:  1.6.1      
 Keywords:  genfromtxt  |  
------------------------+---------------------------------------------------
 I've got installed on one machine (debian linux 6) numpy in 32bit, with
 virtual environment, and on the other machine is numpy in 64bit installed
 via virtual environment.

 There's an issue, if in your file is an float value, but want to read it
 as integer. The example will be more apparent.

 On 32bit:
 {{{
 >>> import sys
 >>> sys.maxint
 2147483647
 >>> from StringIO import StringIO
 >>> import numpy as np
 >>> s = StringIO("333 12.5 22")
 >>> data = np.genfromtxt(s, dtype=[('x', float), ('y', int), ('z', int)])
 >>> data
 array((333.0, 12, 22),
       dtype=[('x', '<f8'), ('y', '<i4'), ('z', '<i4')])
 >>> s = StringIO("333 12.5 22")
 >>> data = np.genfromtxt(s, dtype=[('x', float), ('y', long), ('z', int)])
 >>> data
 array((333.0, -1L, 22),
       dtype=[('x', '<f8'), ('y', '<i8'), ('z', '<i4')])
 }}}

 On 64bit:


 {{{
 >>> import sys
 >>> sys.maxint
 9223372036854775807
 >>> from StringIO import StringIO
 >>> import numpy as np
 >>> s = StringIO("333 12.5 22")
 >>> data = np.genfromtxt(s)
 >>> data
 array([ 333. ,   12.5,   22. ])
 >>> s = StringIO("333 12.5 22")
 >>> data = np.genfromtxt(s, dtype=[('x', float), ('y', int), ('z', int)])
 >>> data
 array((333.0, -1, 22),
       dtype=[('x', '<f8'), ('y', '<i8'), ('z', '<i8')])
 >>> s = StringIO("333 12.5 22")
 >>> data = np.genfromtxt(s, dtype=[('x', float), ('y', long), ('z', int)])
 >>> data
 array((333.0, 12L, 22),
       dtype=[('x', '<f8'), ('y', '<i8'), ('z', '<i8')])
 }}}

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/2037>
NumPy <http://projects.scipy.org/numpy>
My example project