3492bis

Frank Ellermann hmdmhdfmhdjmzdtjmzdtzktdkztdjz at gmail.com
Thu Jun 19 21:13:00 CEST 2008


John C Klensin wrote:

> My preference would be to finish IDNA2008

That would be fine, when somebody asked what "A-label"
is supposed to mean in the IDN test TLD (2606bis) I-D
he had a *big* point.

> _then_ go back and look at Stringprep/3942 (and 
> SASLPrep, etc.) and ask, not just "what is in the
> errata" and "can we move this forward procedurally"
> but, "what did we learn with IDNA that should be
> applied here".

"Nobody is going to invent a new encoding" is in the
WG Charter, I take that as given.  But I forgot the
(in essence ready) 3492bis draft in the discussion
about a "todo" list.  IIRC we talked about it on the
EAI list more than a year ago.

I stumbled about it while trying to reconstruct how
the old IDN WG starting with UTF-5 arrived at 3492 -
meanwhile I found some of the missing pieces, e.g.,
in <http://www.inter-locale.com/IUC22.pdf>, or in
<http://www.nicemice.net/idn/ace-eval.gz>.  I've 
now added a 3492bis link to the Punycode article:
http://en.wikipedia.org/wiki/Punycode#External_links

> "yes, but we need to add the Unicode 5.1 diffs
> and cycles at Proposed"

Likely.  From my POV RFC 3492 is just an algorithm
with input = a sequence of whole numbers between 0
and 0x10FFFF, and output = string of US-ASCII LDH 
characters (not necessarily related to any IDNAbis
purposes).  In theory this could be registered as
charset, or rather, the technical reasons why such
an attempt would be rejected are not obvious.

> that this is just not the right time to ask it.

"Couldn't they find a less obscure algorithm" is 
apparently the most FAQ in IDN discussions, now
after "why is it bound to an old Unicode version"
will be soon obsolete.

 Frank



More information about the Idna-update mailing list