[SciPy-user] bug crash

[David M. Cooke]

On June 16, 2004 02:57 pm, eric jones wrote:
> Pearu Peterson wrote:
> >In [1]: from scipy import *
> >In [6]: b=arange(100000000).astype('f')
> >
> >In [7]: b=b*1j
> >Segmentation fault
>
> It would be nice if the failure wasn't so violent...
>
> eric

Yet another reason to switch to numarray...

>>> import numarray
>>> b = numarray.arange(100000000, type=numarray.Float64)
>>> b = b*1j
Traceback (most recent call last):
  File "<stdin>", line 1, in ?
  File "/usr/lib/python2.3/site-packages/numarray/numarraycore.py", line 723, 
in __mul__
    return ufunc.multiply(self, operand)
  File "/usr/lib/python2.3/site-packages/numarray/ufunc.py", line 877, in 
_cache_miss2
    mode, win1, win2, wout, cfunc, ufargs = \
  File "/usr/lib/python2.3/site-packages/numarray/ufunc.py", line 913, in 
_setup
    if out is None: wout = in1.new(outtype[0])
  File "/usr/lib/python2.3/site-packages/numarray/numarraycore.py", line 643, 
in new
    return self.__class__(shape=self._shape, type=type)
MemoryError
>>>

And by the way, arange(1000000).astype('f') is a bad idea: it allocates an 
array of integers, then makes a new array of floats.
You should use arange(1000000, typecode=Float), which allocates one array of 
Floats.

-- 
|>|\/|<
/--------------------------------------------------------------------------\
|David M. Cooke                      http://arbutus.physics.mcmaster.ca/dmc/
|cookedm at physics.mcmaster.ca