LDH-label terminology

[Frank Ellermann]

Eric Brunner-Williams wrote:
 
> If you think the SPF requirement wasn't octet-aligned

No, it is octet aligned, in theory supporting local parts
in mail as octet label(s).  When John mentioned RFC 2673
it reminded me of the fine tuning of 4408 drafts in 2005.

IMHO we can ignore (really) "binary labels" for IDNAbis.
When I wrote "binary" I meant non-ASCII octets, not 2673.

> I don't much care, other than to have suggested that
> the taxonomy of labels of type data could be consistent
> with 2929, and that text labels are a proper subset of
> binary labels.

Using the term "text label" instead of "octet label" can
muddy the water for our purposes here, if readers think
of "text" as "characters in some charset".  

OTOH using "binary label" instead of "octet label" might
be okay it we agree that we're not talking about RFC 2673.

> the 4408 authors ignored anything other than single-byte
> characters, they weren't wrestling with u-labels.

For SPF-EAI it's UTF-8, and "8" limits this fun to octets.
But that's only remotely related to IDNAbis...

 Frank