Ashley Winters wrote:
> $foo{$x}{$y}{$z} = 10;
>
> set P0, P0[P1]
> set P0, P0[P2]
> set P0[P3], P4
>
> That code assumes the KEY* goes away, which I would like.
> Just don't allow more than one keyed parameter.
You've said what I tried to say, exactly.
> I don't see much use for set_p_[is]_*, . . .
Yep, cool.
I guess the problem I have with the approaches proposed so far
is that they suppose a special key type. I don't particularly
care if the op is called set_p_kp_* as long as the second p is
not in fact required to be some special key type. It could be
just any type whose value could be construed (by a compound
type pmc) as an index. In the simple cases, as an integer by
an array pmc and as a string by a hash pmc; in more complex
case, as a vector of indices by a multi-dim array.
Point is, a int/string/array pmc used that way need not
support any different vtable methods than plain ol'
int/string/array.
Right? Or am I overlooking something?
--
JohnDouglasPorter
__________________________________________________
Do You Yahoo!?
Yahoo! Health - Feel better, live better
http://health.yahoo.com
Thread Previous
|
Thread Next