[NumPy-Tickets] [NumPy] #1720: logical_and () handles nan incorrectly

[NumPy Trac]

#1720: logical_and () handles nan incorrectly
---------------------------+------------------------------------------------
 Reporter:  joshuaschmitt  |       Owner:  somebody
     Type:  defect         |      Status:  new     
 Priority:  normal         |   Milestone:  2.0.0   
Component:  Other          |     Version:  1.5.0   
 Keywords:                 |  
---------------------------+------------------------------------------------

Comment(by pv):

 The behavior of `1 and nan` in Python is a language wart: `(1 and foo) is
 foo`, for any object `foo`. I don't think there is a reason to make
 `logical_and` to follow to this behavior.

 The C behavior does not seem to be specified: at least the C99 standard
 does not say anything about the truth value of NaNs. Conversion from NaN
 to integers is explicitly undefined, and I'd suppose the same applies
 booleans. The `if` statement OTOH is defined via:

 > The first substatement is executed if the expression compares unequal to
 0.
 In the else form, the second substatement is executed if the expression
 compares equal to 0.

 `gcc` apparently implements the first check: the truth value of NaN is
 False in this context (because NaN is not unequal to 0). Assuming that the
 truth value is `True` would however make as much sense.

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/1720#comment:3>
NumPy <http://projects.scipy.org/numpy>
My example project