[Numpy-tickets] [NumPy] #623: loadtxt fails with record arrays

NumPy numpy-tickets@scipy....
Fri Apr 4 12:33:30 CDT 2008


#623: loadtxt fails with record arrays
------------------------+---------------------------------------------------
 Reporter:  mstarzyk    |        Owner:  somebody
     Type:  defect      |       Status:  new     
 Priority:  normal      |    Milestone:  1.0.5   
Component:  numpy.core  |      Version:          
 Severity:  normal      |   Resolution:          
 Keywords:              |  
------------------------+---------------------------------------------------
Comment (by bsouthey):

 With 1.0.5.dev4961, this fails if incorrect dtype of a float instead of an
 int is being used whereas NumPy array() does not fail. Based on the test
 and array wiki documentation:

 {{{
 import numpy as np
 import StringIO

 d2 = StringIO.StringIO()
 d2.write('M 64.0 75.0\nF 25.0 60.0')
 d2.seek(0)

 mydescriptor = {'names': ('gender','age','weight'), 'formats': ('S1',
 'i4', 'f4')}
 b2 = np.array([('M', 64.0, 75.0),('F', 25.0, 60.0)], dtype=mydescriptor)
 print b2
 y2 = np.loadtxt(d2, dtype=mydescriptor)
 }}}
 Gives:

 {{{
 [('M', 64, 75.0) ('F', 25, 60.0)]

     y2 = np.loadtxt(d2, dtype=mydescriptor)
   File "/usr/local/lib64/python2.5/site-packages/numpy/lib/io.py", line
 310, in loadtxt
     row = [converterseq[j](val) for j,val in enumerate(vals)]
 ValueError: invalid literal for int() with base 10: '64.0'

 }}}

 Note that if f4 instead of i4 is used, it works.

-- 
Ticket URL: <http://scipy.org/scipy/numpy/ticket/623#comment:3>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.


More information about the Numpy-tickets mailing list