[perl #76664] [BUG] Null PMC access involving reverse causality when calling a method with s/// syntax in it in Rakudo

[Carl Mäsak via RT]

On Mon Oct 10 16:38:39 2011, coke wrote:
> On Tue Jul 27 18:45:07 2010, coke wrote:
> > On Thu Jul 22 09:58:56 2010, masak wrote:
> > > <masak> rakudo: class A { method pack($_) { s/c// }};
> > > A.new.pack("ccc") # discovered by cono++
> > > <p6eval> rakudo 220b67: OUTPUT«Null PMC access in
> find_method('new')
> > [...]
> > > * masak submits rakudobug
> > >
> > > Remove the s/c//, and everything works. Remove .pack("ccc"), and
> > > everything works. How is it that the .pack call participates in
> > > causing an error related to finding .new ?
> >
> > 21:44 < [Coke]> rakudo: class A { method pack($_) { s/c// }};
> > A.new.pack("ccc")
> >                 # discovered by cono++
> > 21:44 <+p6eval> rakudo 7f5c22: OUTPUT«Cannot modify readonly value␤
> > in
> >                 '&infix:<=>' at line 1␤  in 'A::pack' at line
> >                 22:/tmp/HIyuNvbnYe␤  in main program body at line
> >                 22:/tmp/HIyuNvbnYe␤»
> 
> Pretty much unchanged. Is this closable with tests now?
> 
> 19:37 < [Coke]> rakudo: class A { method pack($_) { s/c// }};
> A.new.pack("ccc")
> 19:37 <+p6eval> rakudo 38907e: OUTPUT«Cannot assign to a readonly
> variable or a
>                 value␤  in sub infix:<=> at src/gen/CORE.setting:7361␤
> in
>                 method pack at /tmp/5h1grLXx6g:1␤  in block <anon> at
>                 /tmp/5h1grLXx6g:1␤  in <anon> at /tmp/5h1grLXx6g:1␤»

It doesn't Null PMC Access any more, so I'm happy. Leaving open in case 
someone decides it needs tests, too.