toplabel (was: A-label definition)

[Frank Ellermann]

John C Klensin wrote:

  [toplabel]
>> RFC 3696 finally got it right, but did not update RFC 1123.
>> Somebody has to fix this, we are in the position to do it.  
 
> 3696 is an informational document that doesn't update
> anything... and can't.

Yes, but as long as nobody did anything about it I needed a
better source than "private communication with John" for the
<toplabel> in various discussions and drafts, and RFC 3696
did the trick.  Ending up in an RFC and an approved PS so far.

> it deliberately takes a very permissive view toward what is
> permitted and has little to do with the present situation,
> much less what is permitted as a TLD name.

It's the situation as with "one dot required" specified in 
RFC 2821 and 3696, later replaced by "maybe not" in 2821bis;
whatever you say is fine, I just need the published source
where you finally stick to one of the possible alternatives.

I'm not interested in a hypothetical TLD ".1-2-3" (covered by
what RFC 3696 and Errata ID 1081 said), "must start with a
letter" as in Errata ID 1353 is plausible and even simpler.

But it's a fairly critical <toplabel> difference from my POV:
<http://www.rfc-editor.org/errata_search.php?eid=1353>
<http://permalink.gmane.org/gmane.ietf.rfc.interest/256>
<http://permalink.gmane.org/gmane.ietf.usenet.format/31982>
<http://permalink.gmane.org/gmane.mail.spam.spf.discuss/24163>
<http://permalink.gmane.org/gmane.ietf.tools/1102>
<http://tools.ietf.org/html/draft-ellermann-idnabis-test-tlds>

It would be really nice if idnabis-rationale can finally nail
this beast, and while at it please decide if "single letter"
is a *technically* possible TLD, or if the ccTLD length is
the minimal length.  This stuff ends up in ABNF and code, it
has to be specified somewhere.  FWIW the USEFOR RFC has "two",
but not "must start with a letter".  And it certainly has no
"updates 1123", idnabis-rationale can pull this stunt PLEASE.

 Frank