[SciPy-dev] A stray pointer somwehere...

[Fernando Perez]

Travis Oliphant wrote:

> Not a bad script for testing long term effects.  I've found the 
> problem.  There were two:  a scalar array object was not being incref'd 
> when retrieved from a.dtype.  Also, and more relevevant to this issue 
> was that .put for OBJECT arrays was not holding on to a reference to the 
> objects placed in the array (nor was it DECREF'ing the current contents).
> 
> This bug is in Numeric as well, but was never discovered -- the put and 
> putmask code came directly from Numeric.

I'm glad to report that your fix checks here as well (on the itanium2 box):

----------------------------------------------------------------------
Ran 134 tests in 0.285s

OK
The tests were run 50 times.

I've attached a slightly fancier version of the test script, in case you want 
to keep it around and occasionally run it.  You can set the verbosity at the 
command line if something out of the ordinary happens.

Thanks for tracking this down so fast!

Cheers,

f
-------------- next part --------------
A non-text attachment was scrubbed...
Name: scipy_test_spin.py
Type: application/x-python
Size: 799 bytes
Desc: not available
URL: <http://mail.python.org/pipermail/scipy-dev/attachments/20051103/c1e184cc/attachment.bin>