M-label definition

[John C Klensin]

Hi.

I'm trying to put notes together for updates to the documents,
even though I don't intend to open them up until Vint makes some
statements about consensus.

I believe that we have agreement that, if the documents are to
discuss mapping at all, an additional term that has been
described as an "M-label" would be useful.   However, in sorting
through my notes, I believe I've seen two different definitions,
i.e.,

	"An M-label is a string that becomes a U-label after
	mapping but it not, itself, a U-label.   In more recent
	terminology, if a U-label is a canonical form, the
	M-label is a non-canonical form of the same thing.  The
	categories of M-label and U-label are disjoint."
	(Patrik and others)
	
	"An M-label is a string that can be mapped into a
	[valid] U-label.  It may be a U-label, since those
	trivially map into themselves.  The category of U-label
	is a proper subset of the category of M-label." (Mark
	and others).

Which one of these fits the text better depends on how the text
is written -- it could be done either way as far as Definitions
and Protocol are concerned, although I think the first
definition might work better for Tables.

I think it would be a really bad (and very confusing) to end up
with M-label(1) and M-label(2) even if different terms are used,
so the WG will need to decide.

    john