[Numpy-tickets] [NumPy] #976: incorrect repr of float32 and float64

NumPy numpy-tickets@scipy....
Tue Dec 23 07:30:30 CST 2008


#976: incorrect repr of float32 and float64
------------------------+---------------------------------------------------
 Reporter:  wxgeo       |       Owner:  somebody
     Type:  defect      |      Status:  new     
 Priority:  normal      |   Milestone:          
Component:  numpy.core  |     Version:  none    
 Severity:  normal      |    Keywords:          
------------------------+---------------------------------------------------
 Hi,
 On my Linux installation, numpy uses comma instead of dot in
 representation of float32, float64 and float96 types (at least) :


 {{{
 >>> import numpy
 >>> numpy.version.version
 '1.2.1'
 >>> numpy.float64(4.3)
 4,2999999999999998
 >>> eval(repr(numpy.float64(4.3)))
 (4, 2999999999999998L)
 }}}

 I use Ubuntu 8.10 (Unbuntu Ibex) kernel 2.6.27-9-generic.
 Default numpy version is 1.1.1, so I compiled 1.2.1, but the same bug
 occurs.

 Not this bug does *not* appear on Windows XP.

 Not also this bug might be french-language relative (although I can't
 actually figure how), since comma is used instead of dot for floating
 number representation in french-speaking countries.

 {{{
 ~$ echo $LANG
 fr_FR.UTF-8
 }}}

-- 
Ticket URL: <http://scipy.org/scipy/numpy/ticket/976>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.


More information about the Numpy-tickets mailing list