(j3.2006) another example of a forward reference

[Van Snyder]

On Fri, 2015-03-13 at 08:52 +0900, Malcolm Cohen wrote:

> Van Snyder remarks:
> <<<
> Or, maybe there isn't a problem, since the type constructor also isn't
> constrained to refer to a previously-defined type.
> >>>
> 
> It is so required, at [81:1].  This is where our habit of inserting notes into 
> the middle of normative text has paid dividends in unreadability.

I had expected this to be a constraint.  Not finding it amongst the
constraints in 4.5.10, I assumed it wasn't prohibited.  But, then, the
requirement on the TYPE specifier isn't a constraint, so why am I not
surprised that [81:1] isn't a constraint?

> Cheers,