[plt-scheme] Dynamic Procedure Invocation

> From: karczma at info.unicaen.fr
> To: plt-scheme at list.cs.brown.edu
> Subject: Re: [plt-scheme] Dynamic Procedure Invocation
> Date: Sat, 26 Mar 2005 10:13:59 +0100
> 
> Robby Findler answers the query: 
> 
> > "Alex Peake" wrote: 
> >> In Common Lisp I can say: 
> >> 
> >> (funcall (find-symbol "+") 1 2 3)
> >> 
> >> -- the key is the proc name is a suitable string (maybe 
> read out of a 
> >> database) --
> >> 
> >> Is there something similar in PLT Scheme (that does not use eval).
>  
> 
> 
> > Without using eval (or procedures beginning with 
> namespace-, that are 
> > essentially eval): no, not in general.
> > 
> > You could make a table of the procedures that you expected 
> to see tho: 
> > 
> > (define table (list (list "+" +) (list "-" -) (list "*" 
> *))) (define 
> > (call sym . args)
> >   (let ([proc (assoc sym table)])
> >     (unless proc
> >       (error 'call "unknown function ~e" sym))
> >     (apply (cadr proc) args)))
> > 
> > (call "+" 1 2 3)
> 
> = = = = = = = = = = = 
> 
> But, essentially, is it really different from the search 
> through the current environment, say,: 
> 
> (define a (string->symbol "+"))
> (define b (namespace-variable-value a))
> (apply b '(12 34 56)) 
> 
> Of course, having a private associative namespace may be much 
> more efficient in searching time, but the namespace handling 
> is (I hope...) quite well optimized. 
> 
> Anyway, this is quite probably what is done in Common Lisp, no? 
> 
> So, why this requirement <<no eval>>? 
> 
> Jerzy Karczmarczuk 
> 

The <<no eval>> is just a performance issue

Alex