From @masak

<jnthn> rakudo​: class A { has $.foo; method new($foo) {
nextwith(​:$foo) } }; say A.new("OH HAI").foo
<p6eval> rakudo 0d4fe0​: OUTPUT«too many named arguments - 'foo' not
expected␤in method A​::new
<masak> jnthn​: should that have worked?
<jnthn> Hmm
<jnthn> rakudo​: class A { has $.foo; method new($foo) { say "ok" } };
say A.new("OH HAI")
<p6eval> rakudo 0d4fe0​: OUTPUT«ok␤1␤»
<jnthn> I think so. I'm not quite sure what it tried to call. :-/
* masak submits rakudobug
<jnthn> rakudo​: class A { has $.foo; method new($foo) { callwith($foo)
} }; say A.new("OH HAI").foo
<p6eval> rakudo 0d4fe0​: OUTPUT«Method 'foo' not found for invocant of
class 'Str'␤»
<jnthn> Oh! :-/
<masak> seems to be something to do with the invocant...
<jnthn> Well
<jnthn> Kinda.
<jnthn> Hmm. I'm not sure off-hand why that doesn't work.
<masak> but it should, so it goes on the RT pile.
<jnthn> One of the annoyances of implementing callwith and nextwith is
that in the case of them being used to defer to another method, they
need to find and pass on the invocant.
<jnthn> Whereas in the case of a wrap, they maybe don't.
<masak> ah. tricky.
<jnthn> Yeah.
<jnthn> It'd not surprise me if we have bugs there.

From @coke

On Sat Aug 15 14​:04​:44 2009, masak wrote​:

<jnthn> rakudo​: class A { has $.foo; method new($foo) {
nextwith(​:$foo) } }; say A.new("OH HAI").foo
<p6eval> rakudo 0d4fe0​: OUTPUT«too many named arguments - 'foo' not
expected␤in method A​::new
<masak> jnthn​: should that have worked?
<jnthn> Hmm
<jnthn> rakudo​: class A { has $.foo; method new($foo) { say "ok" } };
say A.new("OH HAI")
<p6eval> rakudo 0d4fe0​: OUTPUT«ok␤1␤»
<jnthn> I think so. I'm not quite sure what it tried to call. :-/
* masak submits rakudobug
<jnthn> rakudo​: class A { has $.foo; method new($foo) { callwith($foo)
} }; say A.new("OH HAI").foo
<p6eval> rakudo 0d4fe0​: OUTPUT«Method 'foo' not found for invocant of
class 'Str'␤»
<jnthn> Oh! :-/
<masak> seems to be something to do with the invocant...
<jnthn> Well
<jnthn> Kinda.
<jnthn> Hmm. I'm not sure off-hand why that doesn't work.
<masak> but it should, so it goes on the RT pile.
<jnthn> One of the annoyances of implementing callwith and nextwith is
that in the case of them being used to defer to another method, they
need to find and pass on the invocant.
<jnthn> Whereas in the case of a wrap, they maybe don't.
<masak> ah. tricky.
<jnthn> Yeah.
<jnthn> It'd not surprise me if we have bugs there.

Is this the desired behavior?

09​:39 < [Coke]> rakudo​: class A { has $.foo; method new($foo)
{nextwith(​:$foo)
  } }; say A.new("OH HAI").foo
09​:39 <+p6eval> rakudo 142c41​: OUTPUT«OH HAI␤»
09​:39 < [Coke]> rakudo​: class A { has $.foo; method new($foo) { say "ok"
} };
  say A.new("OH HAI")
09​:39 <+p6eval> rakudo 142c41​: OUTPUT«ok␤Bool​::True␤»
09​:39 < [Coke]> rakudo​: class A { has $.foo; method new($foo) {
callwith($foo)
  } }; say A.new("OH HAI").foo
09​:39 <+p6eval> rakudo 142c41​: OUTPUT«Default constructor only takes
named
  arguments␤ in method new at src/gen/CORE.setting​:516␤
in
  <anon> at src/gen/Metamodel.pm​:2375␤ in call_with_args
at
  src/gen/Metamodel.pm​:2372␤ in method new at
/tmp/WW9uUIODIh​:1␤
  in block <anon> at /tmp/WW9uUIODIh​:1␤ in <an…

--
Will "Coke" Coleda

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

From @coke

09​:42 < masak> [Coke]​: looks like the desired behavior.

Closable with tests.

--
Will "Coke" Coleda

From @moritz

Now tested in S12-construction/new.t

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