[Numpy-discussion] Re-implementation of Python Numerical arrays (Numeric) available for download

Peter Verveer Peter.Verveer at embl-heidelberg.de
Mon Nov 19 02:45:02 CST 2001


On Saturday 17 November 2001 00:11 am, you wrote:
> note that my main use of numpy is as a pixel buffer for images. some of
> the changes like avoiding type promotion sounds really good to me :]

I have exactly the same application so I agree with this.

> 7) necessary to add other types?
> yes. i really want unsigned int16 and unsigned int32. all my operations
> are on pixel data, and things can just get messy when i need to treat
> packed color values as signed integers.

Yes please! One of the things that irritates me most on the original Numeric 
is that some types are lacking. I think the whole range of data types should 
be supported, even if some may be seldom used by most people.

> one other thing i'd like there to be a little focus on is adding my own
> new ufunc operators. for image manipulation i'd like new ufunc operators
> that clamp the results to legal values. i'd be happy to do this myself,
> but i don't believe it's possible with the current Numeric.

I write functions in C that directly access the numeric data. I don't use the 
ufunc api. One reason that I do that is that I want my libary of routines to 
be useful independent of  Numeric, so I only have a tiny glue between my C 
routines and Numeric. I hope that it will be still possible to do this in the 
new version.

> the last thing i really really want is for this to be rolled into the
> standard python distribution. that is perhaps the most important aspect
> for me. i do not like requiring the extra dependency for generic numeric
> arrays. :]

I second that!

Cheers, Peter
-- 
Dr. Peter J. Verveer
Bastiaens Group
Cell Biology and Cell Biophysics Programme
EMBL
Meyerhofstrasse 1
D-69117 Heidelberg
Germany
Tel. : +49 6221 387245
Fax  : +49 6221 387242
Email: Peter.Verveer at embl-heidelberg.de





More information about the Numpy-discussion mailing list