Re: [boost] Boost.Multimethod proposal

Subject: Re: [boost] Boost.Multimethod proposal
From: David Bergman (David.Bergman_at_[hidden])
Date: 2009-08-23 13:16:20

On Aug 23, 2009, at 10:00 AM, Mathias Gaunard wrote:

> OvermindDL1 wrote:
>
>> Hmm, I know LISP has a well optimized multimethod implementation
>> written in pure LISP, wonder how easily that method could be
>> transferred to C++, although more verbose...
>
> Lisp is dynamically typed, so the implementation of multimethods is  
> simply that of overloading.

This sentence is a bit oxymoronic. A dynamically typed language has no  
way to overload anything, i.e. overloading is a feature found in  
languages where type can be expressed and dispatched upon.  
Disregarding the oxymoronic quality, why is this implication natural  
(dynamic ---> multimethod == overloading...)?

That said, yes, in CLOS one can extract type information (a tag) and  
match it against the special form arguments provided to a "method" (ad- 
hoc multipolymorphic operator...). So, in that sense, CLOS is not  
completely dynamic, since type information can be expressed statically  
(in these special 'defmethod' forms) and matched against through a  
specialized pattern matching mechanism.

/David