[PEAK] dispatch -- more specific

[Phillip J. Eby]

At 09:56 PM 12/13/2005 +0100, Simon Belak wrote:
>Phillip J. Eby wrote:
>>I'm not sure what usefulness you see in this idea, though, other than 
>>that it allows you to omit the common condition for individual methods.
>
>What I have is basically a glorified type-casting system with two distinct 
>modes of operation. It excepts either a pair (factory, type_wannabe) or 
>(recipe_for_type, type_wannabe) where recipe_for_type can be in several 
>different formats (classes) with default handlers (these are actually 
>causing the problems otherwise it would be just a case of testing for 
>interface compliance) for each branch. Since there will be many cases for 
>each branch I would like to keep repetition down. As a nice bonus this 
>would allow me to refractor the "root" cases if need be.
>
>Do you consider this to be a viable approach (in terms of generic function 
>usage)?

Dunno, I still have no idea what you're actually doing.  :)  I need 
examples, I think.