On Fri, 15 Sep 2017 11:47:10 -0700, c...@zoffix.com wrote:
> On Tue, 05 Sep 2017 03:19:45 -0700, zef...@fysh.org wrote:
> > > (3, Nil)
> > (3 Nil)
> > > List.new(3, Nil)
> > (3 (Any))
> >
> > It is possible (as it should be) to put Nil into a List, if it's
> > constructed using the comma operator. But constructing it via the
> > List.new() constructor replaces the Nil with Any. I think this is a
> > bug
> > in List.new(): it should accept any argument, including Nil, and put
> > it
> > into a List.
> >
> > -zefram
>
>
> Also spotting similar behaviour with Map.new. Since (unlike a Hash),
> it's not creating containers for its things, I'd expect it to retain
> the Nil as Nil.
>
> m: dd Map.new: ("z", Nil)
> rakudo-moar f925c6: OUTPUT: «Map.new((:z(Any)))»
As with #130970 I think this is just a matter of deciding what
slurpy/is_raw behavior each of these should have and updating
docs/spec/code accordingly. Just before doing that, we need to check
on what the roadmap is for some of these types -- are they supposed
to be parameterizable and/or have defaults in the future, or not?
If so, can we specify, rather than just leave it up to conjecture,
that the default parameterization with no default applied will take Nils?
