Elizabeth Mattijsen via RT wrote: >Feels to me that Nil should not be allowed as a type constraint at all.
It's a type. The correct behaviour of it as a type constraint is obvious from that observation, and it does in fact exhibit that correct behaviour. (My "mostly" remark was referring to a problem I thought I was seeing but turned out to be me being misled by [perl #130874], with extra confusion supplied by [perl #130876].) I don't see good grounds to forbid it. Indeed, if it were forbidden then that would be surprising, and you'd have to document how it can't be used in the way that all other types can. -zefram