[Python-Dev] Coercion (RE: Discussion: Introducing new operators
for matrix computation)
Thu, 20 Jul 2000 12:42:44 -0700 (PDT)
On Thu, 20 Jul 2000, Thomas Wouters wrote:
> On Thu, Jul 20, 2000 at 05:49:20PM +0200, M.-A. Lemburg wrote:
> > [Adding new operators]
> > Just a side note before everyone forgets: I think we'll
> > need to carefully redefine coercion *before* adding any
> > new slots/operators/etc. to Python.
> > The current model where coercion is centralized won't
> > scale well with the new operators. My (old) proposal for this
> > (see my Python Pages) was to decentralize coercion and
> > make the coercion process part of the operators task.
> > Thoughts ?
Just a note. The additional operators for numerical computation have
exactly the same properties as the conventional math operators. Their
default on numbers are identical. Don't know if this is relevant, though.
> Agreed. I also think, but that's a personal opinion, that adding new
> operators should be done carefully, and I'm not sure if 2.0 is the right
> time for it. It depends on how far away 2.0 is, of course, not just to make
> sure the operators are implemented right, with clear semantics and
> everything, but also to give everyone a chance to accept the idea of new
> operators ;P