Re: [Chicken-users] current status of tinyclos?

[felix winkelmann]

On 8/12/05, Rick Taube <address@hidden> wrote:

>Am I correct in inferring that Chicken's
> tinyclos has no slot description interface? For example, oop
> implementations in the other three schemes allow for slot initializers,
> keyword initargs, generic accessors and so on. Are these also available
> in chicken? It doesnt seem so from the docs but the text may be
> out-of-date or incomplete. (I certainly could expect so given how
> rapidly chicken and eggs seem to evolve.)  If the support isnt there,
> is it in the queue for a later release or are you waiting for some that
> needs it to implement it? 

Sorry, the tinyclos helper syntax is pretty primitive. You are right, there
isn't anything along slot-descriptions or fancy helper stuff yet.
I haven't got around implementing it.

>Also, I would be interested to find out some
> idea of how "fast" tinyclos method dispatch and slot access is expected
> to be. one of the things we would like to pursue in chicken is realtime
> music io (midi/supercollider) based on srfi-18 and using tinyclos to
> represent music and sound objects. given chickens c connection i am
> hopeful but how realistic is this? Sorry for the newbie questions...

(No need to be sorry.)

Tinyclos is not particularly fast, the best way is to run some test-programs
to measure method-dispatch, depending on your use of OOP features
(do you use multiple inheritance, for example?). If you need a fast
object-system, you can take a look at meroon (which has a more 
developed syntax, generated accessor methods, etc.). On the other
hand meroon is more statical in nature, and doesn't do multiple
inheritance (but multiple dispatch).
I can't recall the exact results, but a simple method-dispatching
benchmark showed meroon (single-dispatch) method calls were
about 2 times slower than normal function calls. Accessors are also
much more efficient than in tinyclos.


cheers,
felix