[SciPy-user] Arrayfns in numpy?

Fernando Perez fperez.net@gmail....
Wed Mar 7 02:12:37 CST 2007


On 3/7/07, Robert Kern <robert.kern@gmail.com> wrote:

> > -1.  If they are going to go in, make them first-class citizens.
>
> Except for interp, those functions really don't deserve to be first class. I
> can't even understand what most of them do on a first (and second and third,
> frankly) pass through their comments.

Well, then maybe they just don't belong in the code at all if they are so bad?

My take on it is just a simple one: if it's functionally unique, it
shouldn't be in oldnumeric because then people will pull it out of
there forever, because no amount of 'numpyfying' their code will give
them that particular feature, since it's not an API issue.

Hopefully oldnumeric is a module whose usage should decrease with
time, and I'm trying to avoid something that would prevent that from
being possible.

But that's why I qualified my statement with '*if* they are going in'.
 Perhaps the right solution is for them NOT to go in, if they are
really bad enough.

I'll shut up now, ultimately I won't be using this anyway :)

Cheers,

f


More information about the SciPy-user mailing list