[Numpy-discussion] About NumPy description

Francesc Altet faltet@carabos....
Tue Apr 3 10:47:22 CDT 2007


I've seen the description of NumPy in the recent announcement (btw, good 
work!) and I think it misses something important. To put this in context, let 
me paste the description used for the latest announce:

NumPy is a Python extension that provides a multi-dimensional array and 
basic mathematical processing to Python.  NumPy also provides foundation 
for basic image and signal processing, linear algebra and Fourier 
transforms, and random number generation.  NumPy also includes easy 
integration with ctypes and Fortran allowing powerful extensions to be 

This is ok but I think it stresses a bit too much the mathematical side of 
NumPy and doesn't mention the important field of application for the flexible 
multi-dimensional data containers that it provides. Namely, these containers 
does offer support for completely general datatypes (not only numerical 
types, but also ascii strings and unicode strings)  that can be combined in a 
general way (even forming heterogeneous and/or nested datatypes) in one 
single container in a very efficient manner (both in terms of speed and 

I think that mentioning this concept can bring the attention of more people, 
and most specially, the developers of database wrappers. One should not 
forget that one of the limitations of the database interfaces in Python is 
that each element retrieved from the database has to be wrapped with a Python 
container (normally string or int/float objects), and this is very ineficient 
when you have to retrieve large datasets from databases. IMHO, database 
developers should learn about this capability of NumPy for the sake of 
achieving efficient databases interfaces for Python.


>0,0<   Francesc Altet     http://www.carabos.com/
V   V   Cárabos Coop. V.   Enjoy Data

More information about the Numpy-discussion mailing list