From @masak

<nebuchadnezzar> hi
<nebuchadnezzar> looking at perl6 book, I tried the examples and find
something curious​: http://pastebin.com/v835G1MD
<nebuchadnezzar> "given" in the sub does not seems to behave the same
way as outside
* masak is looking into it now
<masak> reproduced the bug locally.
<masak> golfing.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
sub p($a, $b) { w $a, $b }; say p(R, R); say w(R, R)
<p6eval> ..rakudo 7e1b38​: OUTPUT«-1␤0␤»
* masak submits rakudobug
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
(-> $a, $b { say w $a, $b })(R, R); say w R, R
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤0␤»
<masak> correction​: the parameter binding is messing it up somehow.
<jnthn> masak​: Was the issue there with star?
<masak> star​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
(-> $a, $b { say w $a, $b })(R, R); say w R, R
<p6eval> star 2012.07​: OUTPUT«-1␤0␤»
<masak> yep.
<jnthn> OK, not a qast related regression then.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; my
($a, $b) = (-> $a, $b { $a, $b })(R, R); say w $a, $b
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> curiouser and curiouser.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; my
($a, $b) = R, R; say w $a, $b
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> a-ha!
<masak> not parameter binding. variables!
<masak> containers, most like.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; my
@​x = R, R; say w(|@​x)
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> aye.
<jnthn> Hm
<jnthn> Can you make it happen without type catpures?
<jnthn> Or are they a necesary part of it?
<masak> I have a hunch they are.
<masak> r​: class R {}; multi w(R, R) { 0 }; multi w($, $) { -1 }; my
@​x = R, R; say w(|@​x)
<p6eval> rakudo 7e1b38​: OUTPUT«0␤»
<masak> aye. they are part of it.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
say w(|[R, R])
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> r​: multi w(​::T, T) { 0 }; multi w($, $) { -1 }; say w(|[1, 1])
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> r​: multi w(​::T, T) { 0 }; multi w($, $) { -1 }; say w(1, 1)
<p6eval> rakudo 7e1b38​: OUTPUT«0␤»
* masak loves golfin' :)
<masak> r​: sub w(​::T, T) { 0 }; say w(|[1, 1])
<p6eval> rakudo 7e1b38​: OUTPUT«Nominal type check failed for parameter
''; expected Scalar but got Int instead␤ in sub w [...]
<masak> yay, I tricked Rakudo into telling me what's wrong, even!
<jnthn> masak++ # that makes it rather clearer
* masak does a victory dance

From @jnthn

On Sun Aug 05 08​:18​:18 2012, masak wrote​:

<nebuchadnezzar> hi
<nebuchadnezzar> looking at perl6 book, I tried the examples and find
something curious​: http://pastebin.com/v835G1MD
<nebuchadnezzar> "given" in the sub does not seems to behave the same
way as outside
* masak is looking into it now
<masak> reproduced the bug locally.
<masak> golfing.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
sub p($a, $b) { w $a, $b }; say p(R, R); say w(R, R)
<p6eval> ..rakudo 7e1b38​: OUTPUT«-1␤0␤»
* masak submits rakudobug
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
(-> $a, $b { say w $a, $b })(R, R); say w R, R
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤0␤»
<masak> correction​: the parameter binding is messing it up somehow.
<jnthn> masak​: Was the issue there with star?
<masak> star​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
(-> $a, $b { say w $a, $b })(R, R); say w R, R
<p6eval> star 2012.07​: OUTPUT«-1␤0␤»
<masak> yep.
<jnthn> OK, not a qast related regression then.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; my
($a, $b) = (-> $a, $b { $a, $b })(R, R); say w $a, $b
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> curiouser and curiouser.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; my
($a, $b) = R, R; say w $a, $b
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> a-ha!
<masak> not parameter binding. variables!
<masak> containers, most like.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; my
@​x = R, R; say w(|@​x)
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> aye.
<jnthn> Hm
<jnthn> Can you make it happen without type catpures?
<jnthn> Or are they a necesary part of it?
<masak> I have a hunch they are.
<masak> r​: class R {}; multi w(R, R) { 0 }; multi w($, $) { -1 }; my
@​x = R, R; say w(|@​x)
<p6eval> rakudo 7e1b38​: OUTPUT«0␤»
<masak> aye. they are part of it.
<masak> r​: class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 };
say w(|[R, R])
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> r​: multi w(​::T, T) { 0 }; multi w($, $) { -1 }; say w(|[1, 1])
<p6eval> rakudo 7e1b38​: OUTPUT«-1␤»
<masak> r​: multi w(​::T, T) { 0 }; multi w($, $) { -1 }; say w(1, 1)
<p6eval> rakudo 7e1b38​: OUTPUT«0␤»
* masak loves golfin' :)
<masak> r​: sub w(​::T, T) { 0 }; say w(|[1, 1])
<p6eval> rakudo 7e1b38​: OUTPUT«Nominal type check failed for parameter
''; expected Scalar but got Int instead␤ in sub w [...]
<masak> yay, I tricked Rakudo into telling me what's wrong, even!
<jnthn> masak++ # that makes it rather clearer
* masak does a victory dance

Fixed now​:

sub w(​::T, T) { 0 }; say w(|[1, 1])
0
class R {}; multi w(​::T, T) { 0 }; multi w($, $) { -1 }; sub p($a, $b)
{ w $a, $b }; say p(R, R); say w(R, R)
0
0

Tagging testneeded.

/jnthn

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

From @moritz

Now tested in S06-multi/type-based.t

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