Re: [boost] Runtime Dynamic Dispatch (boost-dispatch)

From: Matthew Herrmann (matthew.herrmann_at_[hidden])
Date: 2006-10-19 00:17:32

>
> Message: 11 Date: Thu, 19 Oct 2006 01:52:32 +0200 From: Janek Kozicki 
> <janek_listy_at_[hidden]> Subject: Re: [boost] Runtime Dynamic Dispatch 
> (boost-dispatch) To: boost_at_[hidden] Message-ID: 
> <20061019015232.42b1ce98_at_absurd> Content-Type: text/plain; 
> charset=US-ASCII Dean Michael Berris said: (by the date of Wed, 18 Oct 
> 2006 22:24:09 +0800)
>> > Hi Everyone,
>> > 
>> > I'm about to make a shameless plug about the runtime dispatch library
>> > I've been working on the past few months (mostly testing and using in
>> > a project), which I've uploaded to the vault:
>> > http://tinyurl.com/ycsq6n for everyone to check out and comment on.
>>     
>
> can it be used for multimethods ?
Hi Michael,

It looks like the rival for your library is not a switch statement or an 
if-else statement, but a hash_map. In your rationale, you should make 
clear what your library can do that the following construct cannot:

    template <typename T>
    struct dynamic_dispatcher
    {
        typedef hash_map< T, boost::function<void ()> > type;
    };

    dynamic_dispatcher<int>::type dispatcher;
    dispatcher[0] = boost::bind(&myfun0);
    dispatcher[1] = boost::bind(&myfun1);
    int i;
    std::cin >> i;
    dispatcher[i]()

 From my reading, the only difference is that hash map's operator[] will 
not throw in the case where an element is not found.

Best Regards,
Matthew Herrmann
Zomojo Pty Ltd