From @masak

<masak> rakudo​: role A { our sub foo { say "OH HAI" } }; A​::foo
<p6eval> rakudo 5ac05e​: OUTPUT«Cannot find sub A​::foo␤ in main
program body at line 1␤»
<masak> jnthn​: should that work?
<jnthn> mumble
* masak submits rakudobug
<jnthn> :)
<jnthn> I don't know, tbh.
<jnthn> I can argue it both ways.
<masak> me too.
<masak> but I wrote it by accident, and now that I'm thinking about
it, it doesn't seem so odd.
<jnthn> role A { our sub foo { say $?CLASS } } # ouch :)
<masak> well, yeah. but there will always be edge cases.
<jnthn> True
<masak> I can distinguish methods using the $obj.A​::foo syntax, so why
not subs using just A​::foo?

From @FROGGS

TimToady​: as far as I can see through my clouded branes, an our sub
oughta be okay just about anywhere, insofar as it's just a global
really. I wouldn't necessarily expect it to instantiate $?CLASS when
composed though.

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

From @jnthn

On Sun Apr 17 03​:17​:40 2011, masak wrote​:

<masak> rakudo​: role A { our sub foo { say "OH HAI" } }; A​::foo
<p6eval> rakudo 5ac05e​: OUTPUT«Cannot find sub A​::foo␤ in main
program body at line 1␤»
<masak> jnthn​: should that work?
<jnthn> mumble
* masak submits rakudobug

Now it's decided it won't be allowed, S14 is updated, and you get a compile time error. Test in S14-roles/namespaced.t.

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