[NumPy-Tickets] [NumPy] #1487: vectorize(): result depends on execution order; is this normal?

[NumPy Trac]

#1487: vectorize(): result depends on execution order; is this normal?
------------------------------+---------------------------------------------
 Reporter:  lebigot           |       Owner:  somebody
     Type:  defect            |      Status:  new     
 Priority:  normal            |   Milestone:  2.0.0   
Component:  numpy.core        |     Version:  1.3.0   
 Keywords:  vectorize, dtype  |  
------------------------------+---------------------------------------------
 The dtype of the array returned by a vectorize()'d function depends on the
 execution order:

 Session 1:
 {{{
 In [1]: f_vec = vectorize(lambda x: x)
 In [3]: f_vec(arange(3))
 Out[3]: array([0, 1, 2])
 In [4]: f_vec(arange(0.1, 3))  # floats
 Out[4]: array([0, 1, 2])  # integers
 }}}
 Session 2:
 {{{
 In [1]: f_vec = vectorize(lambda x: x)
 In [2]: f_vec(arange(0.1, 3))  # floats, same as above
 Out[2]: array([ 0.1,  1.1,  2.1])  # floats, and not integers as above
 }}}

 Thus, vectorize() has a serious side effect: the dtype of an array
 returned by a vectorized function depends on which vectorized function was
 called first!

 This creates bugs that are hard to detect, because such a behavior is
 highly unusual.

 I had understood from the documentation of vectorize(f) that the type of
 the first value returned by '''f''' instead determined the dtype of the
 returned array…

-- 
Ticket URL: <http://projects.scipy.org/numpy/ticket/1487>
NumPy <http://projects.scipy.org/numpy>
My example project