From: Robert Feldt <feldt@...>
Date: 2001-11-30T00:36:58+09:00
Subject: [ruby-talk:26913] Re: overload possible?

On Thu, 29 Nov 2001, ts wrote:

> >>>>> "R" == Robert Feldt <feldt@ce.chalmers.se> writes:
> 
> R> solutions to a "real" overload module without too much hassle. Still don't
> R> know if its something one would use but at least it would let us
> R> experiment with multi-method dispatch to see if it would add something to
> R> Ruby.
> 
>  See  [ruby-talk:14940]
> 
Ok, you don't like them. Is it because it was hard to find formal
semantics for how to dispatch or did you find other problems?

I guess there must be solutions to the dispatch problem since several
languages have this. Or is the problem that its hard to find a
semantics that feel natural/intuitive? 

There is also a paper by Ferragina et al where they
present an efficient algorithm/data structure for solving the dispatch
problem. Probably overkill for an experimental Ruby implementation but you
get near-linear space and O(log log n) time:

http://citeseer.nj.nec.com/ferragina99multimethod.html

Regards,

Robert