[Python-Dev] type categories

[Greg Ewing]

David Abrahams <dave@boost-consulting.com>:

> I think it's unnatural to force users to associate a function
> designed to be considered symmetrically over a combination of types
> ... with a single one of those types.

I agree, but the alternative seems to be for it to reside
"out there" somewhere in an unknown location from the point
of view of code which uses it.

> Are you talking about allowing the "self" argument of a multimethod to
> appear in any position in the argument list?

Something like that. I haven't really thought it through
very much. Maybe the method name could be systematically
modified to indicate which argument position is "self",
or something like that.

> > That doesn't solve the problem, I know, but at least it
> > would be explicit about not solving it!
> 
> And that's progress?

Maybe. I don't know. At least it would generalise and make available
to the user what's already there in an ad-hoc way to deal with numeric
types.

> Class authors had better be able to say precisely
> which module's idea of "some_function" they're implementing. If you want
> class authors to write something like:
> 
>     def some_module.some_function(T1: x, T2: y)

That would only be an issue if T1 and T2 were already both using the
name some_function for incompatible purposes. That can happen now
anyway with multiple inheritance -- name clashes can occur whenever
you inherit from two pre-existing classes. I don't see that it's any
more of a problem here.

Greg Ewing, Computer Science Dept, +--------------------------------------+
University of Canterbury,	   | A citizen of NewZealandCorp, a	  |
Christchurch, New Zealand	   | wholly-owned subsidiary of USA Inc.  |
greg@cosc.canterbury.ac.nz	   +--------------------------------------+