[pypy-issue] [issue1008] numpypy: commit the code for empty_like, zeros_like, ones_like

mikefc tracker at bugs.pypy.org
Thu Jan 19 02:40:38 CET 2012


mikefc <coolbutuseless at gmail.com> added the comment:

Some points for discussion:
* pypy is developed with TDD, so there will need to be test functions along with 
each new function/method.  (See 
pypy/module/test_lib_pypy/numpypy/core/test_fromnumeric.py for some simple tests I 
wrote).
* Are we going to want some of these functions at App level or interp level?  I 
know it's easy to write them at app level, but where do we want to draw the line 
between the two?  Where is the speed of interp level going to matter versus the 
speed of getting code written?
* There may need to be a refactor of the interp level numpypy layout.   
interp_numarray.py is going to get very crowded as more and more is added.  Should 
the numpy layout be mirrored here (as in the app level module)?

----------
nosy: +mikefc

________________________________________
PyPy bug tracker <tracker at bugs.pypy.org>
<https://bugs.pypy.org/issue1008>
________________________________________


More information about the pypy-issue mailing list