[NumPy-Tickets] [NumPy] #1641: Strange behavior of ldexp on Python2.5 + Wine

NumPy Trac numpy-tickets@scipy....
Tue Oct 19 17:13:46 CDT 2010


#1641: Strange behavior of ldexp on Python2.5 + Wine
-------------------------+--------------------------------------------------
  Reporter:  pv          |       Owner:  somebody
      Type:  defect      |      Status:  closed  
  Priority:  normal      |   Milestone:  1.5.1   
 Component:  numpy.core  |     Version:  1.5.0   
Resolution:  fixed       |    Keywords:          
-------------------------+--------------------------------------------------
Changes (by pv):

 * cc: rgommers (added)
  * status:  new => closed
  * resolution:  => fixed


Comment:

 Ok, got WineDBG working (see http://www.winehq.org/pipermail/wine-
 devel/2010-July/084788.html for an important hint).

 The reason seems to be that python.org 2.5 binary is linked with a C
 runtime that has a broken implementation of `ldexp` (oh joy), and this is
 also what Numpy gets. This is easily tested by inserting
 `printf("\n>>>%g<<<<\n", ldexp(2.0, 2147483647));` in e.g.
 `multiarraymodule.c:initmultiarray`.

 We cannot fix this, as it's not our bug. The test should just be marked as
 KNOWNFAIL on Python < 2.5 on Windows. Marked so in commit:3747ab5 and
 commit:c6504f5

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/1641#comment:1>
NumPy <http://projects.scipy.org/numpy>
My example project


More information about the NumPy-Tickets mailing list