[Numpy-discussion] Toward release 1.0 of NumPy

Travis Oliphant oliphant at ee.byu.edu
Thu Apr 13 12:49:02 CDT 2006


Sasha wrote:

>On 4/12/06, Travis Oliphant <oliphant at ee.byu.edu> wrote:
>  
>
>>...        This also dove-tails nicely
>>with the Python 2.5 release schedule so that NumPy 1.0 should work with
>>Python 2.5 and be fully 64-bit capable for handling very-large arrays.
>>
>>    
>>
>
>I would like to mention one feature that is going to appear in Python
>2.5 that is covering some of the functionality of NumPy.  I am talking
>about the ctypes module
><http://starship.python.net/crew/theller/ctypes/tutorial.html>.  Like
>NumPy, ctypes provides a set of python classes that represent basic C
>types:
>
>        c_byte
>        c_char
>        c_char_p
>        c_double
>        c_float
>        c_int
>        c_long
>        c_short
>        c_ubyte
>         ...
>
>and the ability to describe composite structures.  The later
>functionality is very close to what dtype class provides in numpy.
>
>There are some features in ctype that I like better than similar
>features in numpy.  For example, in ctypes a fixed width array is
>described by multiplying basic type by an integer:
>  
>
>>>>c_char * 10
>>>>        
>>>>
><class '__main__.c_char_Array_10'>
>
>I find this approach more elegant than numpy's dtype('S10').
>
>It looks like there is some synergy to be exploited here, particularly
>in the area of record arrays.
>  
>

Definitely.  I'm not familiar enough with c_types to do this.  Any help 
is appreciated.

-Travis






More information about the Numpy-discussion mailing list