From @masak

$ perl6 -e 'my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say %h{*}[1].perl'
elements() not implemented in class 'Undef'
[...]

When this works, we'll want this to output '[2, 5]'.

From @bbkr

On Kiev build
$ perl6 -e 'my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say %h{*}[1].perl'
Method 'postcircumfix​:<[ ]>' not found for invocant of class 'Proxy'
  in main program body at line 1

The RT System itself - Status changed from 'new' to 'open'

From @diakopter

On Mon Jul 12 06​:51​:06 2010, bbkr wrote​:

On Kiev build
$ perl6 -e 'my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say %h{*}
[1].perl'
Method 'postcircumfix​:<[ ]>' not found for invocant of class 'Proxy'
in main program body at line 1

new behavior​:

17​:13 <diakopter> r​: my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say %h{*}
[1].perl
17​:13 <p6eval> rakudo 024843​: OUTPUT«[4, 5, 6]␤»

From @pmichaud

After thinking about this a bit, I'm inclined to disagree with the
original ticket.

  my %h = 'foo' => [1,2,3], 'bar' => [4,5,6];

If %h{*} is analogous to @​array[*], then it returns a list (slice) of
all of the elements of %h. The .[1] should then return the second
element of this slice, which would be either [1,2,3] or [4,5,6]
depending on the order in which the values were returned from %h{*}.

The real way to grab the second element from each array in the hash
would be something like​:

  %h{*}».[1]

which would apply the .[1] to each array returned from the hash.
Unfortunately this also relies on "duck mapping" of the hyper as defined
in S03, which as yet is NYI in Rakudo. (And implementing S03's duck
mapping will require some refactoring of postcircumfix​:<[ ]> and
postcircumfix​:<{ }>.

To get the (2,5) expected from the original ticket likely requires
something like​:

  > my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say (.[1] for
%h{*}).perl
  (2, 5).list

The other possible mechanism for multidimensional slice subscripting is
defined in S09 using semicolons, as in something like %h{*;1}. However,
that requires declared dimensioned hashes and/or arrays (which the
example is not), and I'm not sure that syntax will work for mixed
hash/array structures.

So, we can either reject this ticket, or we can maybe convert it to a
ticket that notes that "duck mapping" of hypers is NYI in Rakudo
(although I'm not convinced that duck mapped hypers is the correct
solution to the overall problem with hypers).

Pm

From @FROGGS

So far this works​:
m​: my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say %h{*}»[1]
rakudo-moar f185e0​: OUTPUT«5 2␤»

But this does not​:
m​: my %h = 'foo' => [1,2,3], 'bar' => [4,5,6]; say %h{*}».[1]
rakudo-moar f185e0​: OUTPUT«Cannot find method 'postcircumfix​:<( )>'␤ in block <unit> at /tmp/WldouUReun​:1␤␤»

From @moritz

This has been implemented by now, presumably by lizmat.

Tested here​:
Raku/roast@5c4dafa3ab

@moritz - Status changed from 'open' to 'resolved'