From @masak

This be Rakudo db84bc on Parrot r43174.

$ perl6 -e 'class A​::B { my $c; method foo { $c = "OH HAI"; say $c }
}; A​::B.foo'
OH HAI

$ perl6 -e 'class A​::B { my $c; method foo { $A​::B​::c = "OH HAI"; say
$A​::B​::c } }; A​::B.foo'
OH HAI

Fair enough. So far Rakudo agrees with my expectations. But watch this​:

$ perl6 -e 'class A​::B { my $c = 42; method foo { $A​::B​::c = "OH HAI";
say $c } }; A​::B.foo'
42

$ perl6 -e 'class A​::B { my $c = 42; method foo { $c = "OH HAI"; say
$A​::B​::c } }; A​::B.foo'
Use of uninitialized value

Ok, so in the first case we set $A​::B​::c and $c is unaffected. I'd
expect it to change, since it's really the same memory
location/slot/whatever.

In the second case, we set $c and $A​::B​::c is unaffected. Furthermore,
$A​::B​::c isn't even initialized to 42 in this case, even though
$A​::B​::c should only be a longer way of saying $c (inside the A​::B
scope).

From @moritz

On Sun Jan 24 10​:01​:53 2010, masak wrote​:

This be Rakudo db84bc on Parrot r43174.

$ perl6 -e 'class A​::B { my $c; method foo { $c = "OH HAI"; say $c }
}; A​::B.foo'
OH HAI

$ perl6 -e 'class A​::B { my $c; method foo { $A​::B​::c = "OH HAI"; say
$A​::B​::c } }; A​::B.foo'
OH HAI

Fair enough. So far Rakudo agrees with my expectations. But watch this​:

$ perl6 -e 'class A​::B { my $c = 42; method foo { $A​::B​::c = "OH HAI";
say $c } }; A​::B.foo'
42

Now says

Null PMC access in getprop()
current instr.​: '&infix​:<=>' pc 16638 (src/builtins/Junction.pir​:228)
called from Sub 'perl6;A;B;foo' pc 202 (EVAL_1​:96)

Not sure that's an improvement :-)

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

From @bbkr

NOM​: still broken

bbkr​:nom bbkr$ ./perl6 -e 'class A​::B { my $c; method foo { $c = "OH
HAI"; say $c } }; A​::B.foo'
OH HAI

bbkr​:nom bbkr$ ./perl6 -e 'class A​::B { my $c; method foo { $A​::B​::c =
"OH HAI"; say $A​::B​::c } }; A​::B.foo'
OH HAI

bbkr​:nom bbkr$ ./perl6 -e 'class A​::B { my $c = 42; method foo {
$A​::B​::c = "OH HAI"; say $c } }; A​::B.foo'
42

From @masak

<masak> r​: class A​::B { my $c = 42; method foo { $A​::B​::c = "OH HAI"; say
$c } }; A​::B.foo
<p6eval> rakudo 0eea68​: OUTPUT«42␤»
<masak> r​: class A​::B { my $c = 42; method foo { $c = "OH HAI"; say
$A​::B​::c } }; A​::B.foo
<p6eval> rakudo 0eea68​: OUTPUT«Any()␤»
<masak> the former one is fine, I guess. at least if package stashes can
be extended like that.
<masak> the latter one is, hm, also fine.
* masak rejects https://rt-archive.perl.org/perl6/Ticket/Display.html?id=72326

The original ticket seems to stem from a confusion between my-scoped and
our-scoped variables, and the containers they share. I see nothing
unexpected in the above. Rejecting ticket.

@masak - Status changed from 'open' to 'rejected'