On Jul 21, 2007, at 12:42 AM, Marcus Denker wrote:
> What Ted suggests is to use a special "selector name" object (or an
> object encodig
> Selector + Arguments) instead of symbols for #perform and then search
> *the complete system* for this on "senders of". This them will catch
> Selectors that
> are not referenced from a method, but stored in an iVar in some
> object.
>
> For Squeak, this would mean that this would be not allowes:
>
> self perform: #hallo
>
> but this instead:
>
> sel := Selector named: #hallo.
> self perform: sel.
I am not sure what the original proposal would gain, because typically
you do not do a
self perform:#hallo.
where #hallo is any constant/literal selector, because if it were
constant/literal, then you could just write the message-send naturally:
self hallo.
However, the fact that you need to use #perform: when you want to have
a variable selector has bothered me for some time, because it is a
weird asymmetry that, for example, functional programming languages do
not seem to share: if you get passed a function (function pointer in
C and the like), you can call it using natural syntax. So what I
would like to see would be some sort of "selector variable", a
variable that can be used in a standard message expression but is
actually variable:
| variableSelector:withArg: |
variableSelector:withArg: := self computeSelector.
" probably
need to check arity for compatibility here "
self variableSelector:'hello' withArg:2.
Not sure this is workable/agreeable in a Smalltalk context, but
something like it would be really nice. IMHO.
Marcel
_______________________________________________
fonc mailing list
[email protected]
http://vpri.org/mailman/listinfo/fonc
