Ingo Blechschmidt <iblech <at> web.de> writes:
> > then it has a better chance of working, presuming someone has the
> > gumption to write .pick on hashes, which doesn't look entirely trivial
> > to do right.
>
> <thinking out loud>I'm sure I overlooked something, but the following
> seems to be correct and is not *that* difficult :):
> class Hash;
> ...;
> method pick() is rw {
> # First pick a random key.
> my $key = .keys.pick;
> # Then return an appropriate Proxy object:
> return new Proxy:
> FETCH => {
>
> Ok. While typing the C<{> here, I realized you were correct :)
> It'd be reasonable simple if there was a .get_pair_by_key method
> (which'd do appropriate binding and'd be C<is rw>):
>
> method pick() is rw {
> my $key = .keys.pick;
> my $pair := .get_pair_by_key($key);
> return $pair;
> }
well :)
# Ignoring multi-dimensionality
method pairbykey($key is copy) is rw {
return new Proxy:
FETCH => -> $pair { $self{$key} },
STORE => -> $pair {
my ($newkey, $newval) = $pair.kv;
$self.delete($key); # Delete old entry
$self{$key = $newkey} = $newval; # Add new entry
};
}
# Correct?
BTW, this would make the following work, too:
for %hash.pairs -> $pair is rw { $pair = ... }
# Implementation of Hash::pairs:
method pairs (Any|Junction [EMAIL PROTECTED]) {
my @keys = @keytests ?? .keys.grep:{ $_ ~~ any @keytests } :: .keys;
return .pairbykeyÂ.(@keys);
}
# Correct?
The next logical step to full rw-ness would be a rw .pairs (@keytests
removed for simplicity):
method pairs ($self:) is rw {
return new Proxy:
FETCH => { .pairbykeyÂ.(.keys) },
STORE => -> [EMAIL PROTECTED] {
.delete($_) for .keys;
# [Is there a .clear? The obvious way, %hash = ()/
# $self = {} won't work of course.]
$self{$_.key} = $_.value for @new;
};
}
# Correct?
</thinking out loud>
--Ingo
--
Linux, the choice of a GNU | Mr. Cole's Axiom: The sum of the
generation on a dual AMD | intelligence on the planet is a constant;
Athlon! | the population is growing.
