From @moritz

In the course of the 'ng' branch, Rakudo has regressed on passing parent
attributes to constructors.

Tests can be found in t/spec/S12-construction/new.t

@coke - Status changed from 'new' to 'open'

From @masak

<masak> rakudo​: class A { has

$.x; method foo { say $

$.x; method bar { say $

From @masak

<masak> rakudo​: class A { has

$.val }; class B is A { has $

$.val }; class B is A { has $

alpha didn't get the accessor right, but it parses the constructor
syntax without a problem. There are spectests for this in
S12-construction/new.t, but they're TODO'd.

From @coke

On Thu Oct 28 14​:41​:06 2010, masak wrote​:

<masak> rakudo​: class A { has

Extra open brace or missing close brace

$.x; method foo { say $

!x } }; class B
is A { has

Extra open brace or missing close brace

$.x; method bar { say $

!x } }; B.new(​:x(42), A{ :x(5) })
<p6eval> rakudo 479650​: OUTPUT«flattened parameters must be a hash or
array [...]
<masak> jnthn​: whoz op with that?
<jnthn> rakudo​: class A { }; class B { }; B.new(A{ :x(5) })
<p6eval> rakudo 479650​: OUTPUT«flattened parameters must be a hash or
array [...]
<jnthn> Hm
<jnthn> masak​: Off hand, I'm not sure.
<jnthn> masak​: Unless it's a general WHENCE issue.
<masak> but it's a bug, right?
<jnthn> Yeah
* masak submits rakudobug
<jnthn> I wonder if we're missing A{ ... } support somehow.
<masak> I'm pretty sure it worked at some point.
<masak> not sure it ever worked in master, though.

This now dies with​:

Default constructor for 'B' only takes named arguments

Is this acceptable?

--
Will "Coke" Coleda

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

From @masak

On Wed, Feb 26, 2014 at 1​:50 AM, Will Coleda via RT
<perl6-bugs-followup@​perl.org> wrote​:

On Thu Oct 28 14​:41​:06 2010, masak wrote​:

<masak> rakudo​: class A { has

Extra open brace or missing close brace

$.x; method foo { say $

!x } }; class B
is A { has

Extra open brace or missing close brace

$.x; method bar { say $

!x } }; B.new(​:x(42), A{ :x(5) })
<p6eval> rakudo 479650​: OUTPUT«flattened parameters must be a hash or
array [...]
<masak> jnthn​: whoz op with that?
<jnthn> rakudo​: class A { }; class B { }; B.new(A{ :x(5) })
<p6eval> rakudo 479650​: OUTPUT«flattened parameters must be a hash or
array [...]
<jnthn> Hm
<jnthn> masak​: Off hand, I'm not sure.
<jnthn> masak​: Unless it's a general WHENCE issue.
<masak> but it's a bug, right?
<jnthn> Yeah
* masak submits rakudobug
<jnthn> I wonder if we're missing A{ ... } support somehow.
<masak> I'm pretty sure it worked at some point.
<masak> not sure it ever worked in master, though.

This now dies with​:

Default constructor for 'B' only takes named arguments

Is this acceptable?

It shouldn't die, it should be a legal way to construct a B object
with an assignment to the (shadowed) $!x in A.

From @FROGGS

class Foo { has $.x is rw }; class Bar is Foo { }; my Foo $u .= new(​:5x); say $u.x; $u = Bar.new(Foo{​:12x});
rakudo-moar a34d02​: OUTPUT«5␤Default constructor for 'Bar' only takes named arguments␤ in block <unit> at /tmp/FrBrn8xOI_​:1␤␤»

I guess this needs to translate to something else than Any in order to make it work​:
class Foo { }; say (Foo{​:12x})
rakudo-moar a34d02​: OUTPUT«(Any)␤»

From @jnthn

On Thu May 07 02​:51​:39 2015, FROGGS.de wrote​:

class Foo { has $.x is rw }; class Bar is Foo { }; my Foo $u .=
new(​:5x); say $u.x; $u = Bar.new(Foo{​:12x});
rakudo-moar a34d02​: OUTPUT«5␤Default constructor for 'Bar' only takes
named arguments␤ in block <unit> at /tmp/FrBrn8xOI_​:1␤␤»

I guess this needs to translate to something else than Any in order to
make it work​:
class Foo { }; say (Foo{​:12x})
rakudo-moar a34d02​: OUTPUT«(Any)␤»

We actually weren't recognizing the auto-vivifying closure syntax at all. I've now reserved it, so we can implement this feature in a future version of the Perl 6 language. It's not being missed too hard, and I don't want to rush in a half-baked implementation of it for 6.Christmas.