Ian G wrote:
The architectural imperative here is that the definition has to stop
somewhere.
I think there are many architectural imperatives here. "get the
rfc out" is one, "don't spawn non-interoperable implementations" is
another.
"the definition has to stop somewhere" I would have said as
"we wish the definition to be as precise as we can so as to ensure
the definition stops somewhere", but, yeah, that too...