[Numpy-discussion] Status of Numeric

Arthur ajsiegel at optonline.com
Wed Jan 21 21:20:17 CET 2004


On Wed, 21 Jan 2004 13:31:01 -0500, "Humpty Dumpty"
<oliver.schoenborn at utoronto.ca> wrote:

>
>"Paul Dubois" <paul at pfdubois.com> wrote in message
>news:mailman.533.1074574795.12720.python-list at python.org...
>> Having two array classes around is a terrible idea. Once you have two
>> popular array classes the entire world bifurcates. This plot package
>
>I second that vehemently.
>
>> I only suggested we do numarray because Numeric was, by general
>> agreement, unmaintainable.  It was hard to work on and lacked the kind
>> of robustness necessary to get permission to put it in the core. We all
>> agreed Numeric would cease. Let's stick to the plan.
>
>However, I think experience has shown that users tend to go for better
>performance rather than better design.
>
>So why couldn't Numeric be redesigned for maintainability, but keep the
>algorithms that make it more performant than numarray?  Does the interface
>need to change?
>
>Oliver
>

>From where I sit, these look like very important questions in regard
to Python's future.

I certainly have no answers.

Paul seems to mention that one of the goals of numarray was to achieve
something clean and maintainable enough to move to a module of the
Python core distrtibution..  

Which, to me, seems like a worthy goal. 

On the other hand, it would seem that the goal of something to move
into the core would be performance optimized at the range of array
size most commonly encountered.  Rather than for the extraodrinary,
which seems to be the goal of numarray, responding to specific needs
of the numarray development team's applications.

Has the core Python development team given out clues about their
feelings/requirements for a move of either Numeric or numarray into
the core? 

It concerns me that this thread isn't trafficked.

Is it that folks feel that the stauts quo: an actively maintained 3rd
party package for numerical operations on multidimensional arrays, is
adequate.  With numarray to replace Numeric as the standard. And
performance nuances relatively unimportant.

I see nothing terrible about that position, and hope that that is the
tacit measage of the lack of respoonse to the thread.

Certainly folks can't be considering the global issues here simply
unimportant.

Art



More information about the Python-list mailing list