[perl #120071] Losing type info in the bowels of MapIter

[via RT]

# New Ticket Created by  Elizabeth Mattijsen 
# Please include the string:  [perl #120071]
# in the subject line of all future correspondence about this issue. 
# <URL: https://rt.perl.org:443/rt3/Ticket/Display.html?id=120071 >


A deleted entry in an array, returns it type object correctly when directly 
invoked, but a generic Any object when invoked through map.  At least on 
parakudo, not sure about jakudo.


[14:23:32] <lizmat>      r: my Int @a=^3; @a[1]:delete; @a.map:{say .WHAT.perl 
}; @a[1].WHAT.perl.say  # losing type info deep in the bowels of MapIter
[14:23:34] <camelia>     rakudo 0808ac: OUTPUT«Int␤Any␤Int␤Int␤»
[14:24:30] <lizmat>      no more time to look at this now, maybe someone else 
has inspiration

[21:40:20] <lizmat_>     r: my Int @a=^3; @a[1]:delete; @a.map:{say .WHAT.perl 
}; @a[1].WHAT.perl.say  # losing type info deep in the bowels of MapIter
[21:40:48] <+camelia>    rakudo 0808ac: OUTPUT«Int␤Any␤Int␤Int␤»
[21:41:02] <lizmat_>     I guess I'm going to rakudobug this
[21:41:22] <lizmat_>     unless someone has a suggestion?
[21:43:37] <masak>       was the thing the other day rakudobugged? it was 
similar.
[21:44:19] <lizmat_>     I think I told timotimo that I had an idea where it 
was coming from
[21:44:23] <lizmat_>     but I was wrong
[21:44:31] <lizmat_>     at least, in the short term

[12:06:07] <lizmat>      Could someone run this on jakudo and tell me what it 
says?  my Int @a=^3; @a[1]:delete; @a.map:{ .WHAT.perl.say }; 
@a[1].WHAT.perl.say
[12:06:33] <timotimo>    nqp/master + rakudo/nom?
[12:06:56] <lizmat>      doesn't really matter
[12:07:15] <lizmat>      this is deep in MapIter, and that has specific 
versions for Parrot and JVM
[12:08:36] <jnthn>       lizmat:
[12:08:37] <jnthn>       Int
[12:08:40] <jnthn>       Any
[12:08:40] <jnthn>       Int
[12:08:40] <jnthn>       Int
[12:08:50] <timotimo>    that's ... unfortunate
[12:08:51] <lizmat>      ok, so same pb on jakudo
[12:08:56] <jnthn>       (that's on cursless branch, but no MapIter changes 
there)
[12:09:01] <lizmat>      indeed
[12:09:09] <lizmat>      so the pb is in the non-VM specific code
[12:09:16] <lizmat>      s/is/could be/