From @masak

<masak> rakudo​: my @​a = <a b c>; @​a[0 ..^ *-1] >>~=>> "\n"; say @​a.perl
<p6eval> rakudo 015d77​: OUTPUT«(timeout)»
* masak submits rakudobug
<colomon> rakudobug for that?
<masak> colomon​: locally, I get a stack overflow.
<colomon> but do you think it is correct?
<colomon> you're mixing positional whatever with range whatever
<masak> oh, I can't do that?
<colomon> probably shouldn't stack overflow, but...
<masak> I totally thought I could.
<masak> it feels so natural.
<masak> rakudo​: my @​a = <a b c>; @​a[0 ..^ *-1]
<p6eval> rakudo 015d77​: OUTPUT«(timeout)»
<colomon> well, do you know if it is (0 ..^ *) - 1 or 0 ..^ (*-1) (for
starters)?
<masak> I'm assuming the latter, for precedence reasons.
<colomon> and if it is the later, you're making a Range with an Int on
one side and a closure on the other
<masak> so?
<masak> that's what WhateverCode is; a closure.
<masak> I don't see the fundamental difference to ordinary ranging
with Whatever.
<colomon> ranges don't participate in WhateverCode, do they?
<masak> I don't know.
<colomon> Certainly 1..* does not generate a closure
<masak> S09​:245 has @​array[0 .. *-1] as an example of valid slicings.
<colomon> (It does look like it should parse as 0..(*-1))
<jnthn> masak​: In [0..*-1] the *-1 would...yes, that. :)
<masak> jnthn​: that's what I expected.
<masak> or rather, I don't care much how it works implementation-wise
(as long as it works), but that was the precedence I assumed.
<colomon> "For any prefix, postfix, or infix operator that would be
curried by a Whatever, a WhateverCode also autocurries it"
<colomon> to me, that pretty explicitly leaves out the 0..*-1 case
<masak> colomon​: even though S09 says it should work?
<colomon> masak​: the spec is known to have errors
<masak> colomon​: but it would *suck* if this didn't work!
<masak> colomon​: and TimToady wrote that part of S09, so it's not like
it's untrustworthy because of many chefs.
<colomon> personally, I'm not seeing any reason to have the above
quote... it seems like anything hit with a WhateverCode could curry.
<colomon> masak​: TimToady++ is a genius, but he does make mistakes.
<masak> colomon​: well, what I'm saying doesn't depend on TimToady not
making mistakes. it depends on what feels right from a user
perspective.
<masak> and if I won't be able to write [0 .. *-1], I'll fork and make
my own Perl 6.
<jnthn> .oO( Who'd be less tormenting to the implementer... )
<colomon> Look, I don't see any obvious harm in allowing that. But
presumably the specquote is there for a reason which I do not
understand.

From @masak

<masak> perl6​: my @​e = 1, 2, 3; .say for @​e[0 ..^ *]
<p6eval> rakudo 922500, niecza v7-44-g54b5188​: OUTPUT«(timeout)»
<p6eval> ..pugs​: OUTPUT«pugs​: out of memory (requested 1048576 bytes)␤»
<masak> for something I think should work, the implementations seem
unusually united against me.
* masak goes to read S09
<moritz> nom​: my @​e = 1, 2, 3; .say for @​e[0 ..^ *]
<p6eval> nom​: OUTPUT«1␤2␤3␤»
* masak hugs nom
<masak> ok, S09 supports my belief that @​e[0 ..^ *] shouldn't loop infinitely.
* masak submits rakudobug

Specifically, S09​:675 talks about cases like this.

Oh, and as evidenced by moritz++' evaluation, this already works in nom.

From @moritz

works now, and tested in t/spec/S02-types/array.t

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