[Numpy-tickets] [NumPy] #716: Inconsistent behavior, i686 vs x86_64

[NumPy]

#716: Inconsistent behavior, i686 vs x86_64
---------------------------------+------------------------------------------
 Reporter:  jamie@goonathon.net  |        Owner:  somebody
     Type:  defect               |       Status:  closed  
 Priority:  normal               |    Milestone:  1.0.5   
Component:  Other                |      Version:  none    
 Severity:  normal               |   Resolution:  wontfix 
 Keywords:                       |  
---------------------------------+------------------------------------------
Changes (by oliphant):

  * status:  new => closed
  * resolution:  => wontfix

Comment:

 Good example of the draw-backs of automatically upcasting the sum.
 However, it was decided when this was debated, that this result was
 acceptable.

 The only thought that I think is possible is to have .sum() return
 numpy.uint32 by default (for unsigned inputs) even on 64-bit platforms.
 But, I'm not really in favor of this idea either.

 Suggestion is to use the dtype argument to sum to change the reduction
 variable and not rely on the default if you are concerned.

 Division in Python is moving toward "true" division anyway which would
 make both of these cases return float.

-- 
Ticket URL: <http://scipy.org/scipy/numpy/ticket/716#comment:1>
NumPy <http://projects.scipy.org/scipy/numpy>
The fundamental package needed for scientific computing with Python.