From @bbkr

Please check following example​:

[14​:27] <bbkr> rakudo​: my $.foo; say "alive"; $.foo
[14​:27] <p6eval> rakudo : OUTPUT«aliveâ�¤Null PMC access in find_method('foo')â�¤
in main program body at line 22​:/tmp/B60vOmq2flâ�¤Â»

Declaration of class attribute works in main scope. But accessing it does not.

Either appearance of $.foo in main scope should give error about useless
usage/declaration (for example like when you try to write "method foo {}" in
main scope).
This requires STD change I think, because now this syntax is legal.

Or...

Accessing $.foo should not give Null PMC access.

I don't know which one is correct (spec says nothing about class attributes
used in main scope).

From @coke

On Tue Nov 30 05​:36​:43 2010, pawel.pabian@​implix.com wrote​:

Please check following example​:

[14​:27] <bbkr> rakudo​: my $.foo; say "alive"; $.foo
[14​:27] <p6eval> rakudo : OUTPUT«aliveâ�¤Null PMC access in
find_method('foo')�
in main program body at line 22​:/tmp/B60vOmq2flâ�¤Â»

Declaration of class attribute works in main scope. But accessing it
does not.

Either appearance of $.foo in main scope should give error about
useless
usage/declaration (for example like when you try to write "method foo
{}" in
main scope).
This requires STD change I think, because now this syntax is legal.

Or...

Accessing $.foo should not give Null PMC access.

I don't know which one is correct (spec says nothing about class
attributes
used in main scope).

It now warns on the first statement, but proceeds to NPA on the last​:

09​:31 < [Coke]> rakudo​: my $.foo; say "alive"; $.foo # RT 80008
09​:32 <+p6eval> rakudo ce5cca​: OUTPUT«Useless declaration of a has-scoped
  method in mainlineâ�¤aliveâ�¤Null PMC access in find_method('foo')â�¤
  in block <anon> at /tmp/SJaI203Lww​:1â�¤ in <anon> at
  /tmp/SJaI203Lww​:1â�¤Â»

--
Will "Coke" Coleda

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

From @moritz

It now warns about the first, and dies with "Variable $.foo used where
no 'self' is availalbe" for the second, which I think is good enough to
close the ticket. The latter is tested in S32-exceptions/misc.t

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