From @masak

<diakopter> rakudo​: my :​:byte $a
<p6eval> rakudo 70667a​: OUTPUT«Null PMC access in type() [...]
<diakopter> rakudo​: my :​:Byte $a
<p6eval> rakudo 70667a​: OUTPUT«Null PMC access in type() [...]
<masak> I think I can rakudobug those.
<masak> rakudo​: my :​:a $a
<p6eval> rakudo 70667a​: OUTPUT«Null PMC access in type() [...]
* masak submits rakuodbug

From @masak

<diakopter> rakudo​: my :​:foo $x; say $x
<p6eval> rakudo a54677​: OUTPUT«Null PMC access in type() [...]
* masak submits rakudobug
<masak> a phantom-typed variable!
<diakopter> uh oh.
<diakopter> masak​: you're encouraging my rakudofuzzing
<masak> of course!
<jnthn> I'm pretty sure :​:whatever exploding is already in there.
<masak> there are related tickets, for sure​:
http://rt.perl.org/rt3/Ticket/Display.html?id=64642 is one.
<masak> but I find none exactly like this. so I'm submitting it.
<jnthn> I've no idea what we're supposed to do in such cases
<masak> fail at compile-time? ignore the typing? let demons fly out of
the user's nose?
<jnthn> I'd prefer a Null PMC Access to my nose demons escaping.
<jnthn> Maybe the spec says what to do somewhere. :-)
<masak> maybe.

From @masak

<jnthn> rakudo​: my :​:foo $x;
<p6eval> rakudo a54677​: OUTPUT«Null PMC access in type() [...]
<jnthn> masak​: You only got a birdie, not a hole in one. :P
<masak> jnthn​: will update the ticket.
<diakopter> hee
<jnthn> :-)
<jnthn> masak​: Yeah, I was looking at it thinking "hmm, I know how that code works and I'm
surprised we make it past the declaration" :)

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

From radu_cs85@yahoo.com

Added test in S04-declarations/my.t

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

From @kyleha

This is an automatically generated mail to inform you that tests are now available in t/spec/S04-declarations/my.t

commit 1614779058b7c8f759221f1614ba3fab04992895
Author​: radus <radus@​c213334d-75ef-0310-aa23-eaa082d1ae64>
Date​: Sun Jun 27 09​:20​:19 2010 +0000

  [t/spec] Added test for RT 72814 - Null PMC access when typing a my-declared variable as :​:a in Rakudo.
 
  git-svn-id​: http://svn.pugscode.org/pugs@&#8203;31475 c213334d-75ef-0310-aa23-eaa082d1ae64

diff --git a/t/spec/S04-declarations/my.t b/t/spec/S04-declarations/my.t
index 6f23c04..25145f7 100644
--- a/t/spec/S04-declarations/my.t
+++ b/t/spec/S04-declarations/my.t
@@ -1,7 +1,7 @@
 use v6;
 use Test;
 
-plan 62;
+plan 63;
 
 #L<S04/The Relationship of Blocks and Declarations/"declarations, all
 # lexically scoped declarations are visible"> 
@@ -247,4 +247,11 @@ eval_lives_ok 'my (%h?)', 'my (%h?) lives';
 eval_lives_ok 'my $x = 3; class A { has .$y = $x; }; say A.new.y', 
         'global scoped variables are visible inside class definitions';
 
+#RT #72814
+
+{
+    #?rakudo skip 'RT 72814' 
+    lives_ok {my ::a $a}, 'typing a my-declared variable as ::a works.';
+}
+
 # vim: ft=perl6

From @bbkr

NOM​: mentioned test is already unfudged and passing

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

From @coke

On Wed Jun 09 14​:45​:50 2010, masak wrote​:

<jnthn> rakudo​: my :​:foo $x;
<p6eval> rakudo a54677​: OUTPUT�Null PMC access in type() [...]
<jnthn> masak​: You only got a birdie, not a hole in one. :P
<masak> jnthn​: will update the ticket.
<diakopter> hee
<jnthn> :-)
<jnthn> masak​: Yeah, I was looking at it thinking "hmm, I know how
that code works and I'm
surprised we make it past the declaration" :)

Now they say​:

00​:19 < [Coke]> rakudo​: my :​:foo $x; say $x
00​:19 <+p6eval> rakudo cbdd9b​: OUTPUT«Cannot call a method on type variable foo␤ in
  find_method at src/gen/Metamodel.pm​:261␤ in sub say at
  src/gen/CORE.setting​:4722␤ in sub say at src/gen/CORE.setting​:4729␤
  in <anon> at /tmp/ormgC9k2j3​:1␤ in <anon> at /tmp/ormgC9k2j3​:1␤␤»
00​:19 < [Coke]> rakudo​: my :​:foo $x;
00​:19 <+p6eval> rakudo cbdd9b​: ( no output )
00​:19 < [Coke]> rakudo​: my :​:foo $x; say "alive"
00​:19 <+p6eval> rakudo cbdd9b​: OUTPUT«alive␤»

--
Will "Coke" Coleda

From @usev6

The Null PMC access is gone, so I'm closing this ticket.

The error message has changed slightly​:

$ perl6-m -e 'my :​:foo $x; say $x'
Cannot type check against type variable foo
  in any type_check at src/gen/m-Metamodel.nqp​:297
  in block <unit> at -e​:1

I added a test for this exception to S02-names-vars/names.t with commit Raku/roast@5e0641d092

If you don't think current behaviour is correct, please reopen the ticket.

Also, there was another Ticket (#​72814; with test in S04-declarations/my.t) about the Null PMC access with 'my :​:foo $x;'. I'm gonna merge both tickets.

@usev6 - Status changed from 'open' to 'resolved'

From @coke

On Tue Oct 28 14​:41​:35 2014, bartolin@​gmx.de wrote​:

The Null PMC access is gone, so I'm closing this ticket.

The error message has changed slightly​:

$ perl6-m -e 'my :​:foo $x; say $x'
Cannot type check against type variable foo
in any type_check at src/gen/m-Metamodel.nqp​:297
in block <unit> at -e​:1

I added a test for this exception to S02-names-vars/names.t with
commit Raku/roast@5e0641d092

If you don't think current behaviour is correct, please reopen the
ticket.

Also, there was another Ticket (#​72814; with test in S04-
declarations/my.t) about the Null PMC access with 'my :​:foo $x;'. I'm
gonna merge both tickets.

This test now dies on the JVM with a java.lang.StackOverflowError

re-opening ticket.
--
Will "Coke" Coleda

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

From @usev6

The test passes again (rakudo-j). I'm closing this ticket as resolved.

@usev6 - Status changed from 'open' to 'resolved'