Re: [idn] Why we can go directly to UTF-8

["Eric A. Hall" <ehall@ehsco.com>]

> Those systems enforce LDH, and will reject non-LDH sets.

If we're into lists, here's one I'd like to see: all of the systems which
provide user input of domain names, and which are also aware of the
variable-octets underlying UTF-8 domain names. These systems need to know
that the domain name will have a maximum label length of 63 octets and
prevent additional input when the UTF-8 octet limit is exceeded, which
will be dramatically different than the displayed character count.

The list of systems which do support this will certainly be shorter than
the list of systems which don't, so that is an action item for you I would
guess. :/

Now look, seriously, there is *no* *freakin* *way* that we can go to
just-do-UTF-8. At the very least resolvers need to know about the
difference between wide() and normal names, systems need $HOSTNAME calls
that support UTF-8 encodings, etc. This is a massive undertaking.

I am clearly on record in the archives as supporting a clean break from
the past, but it must be a clean break. Fudgery just will not work.

-- 
Eric A. Hall                                        http://www.ehsco.com/
Internet Core Protocols          http://www.oreilly.com/catalog/coreprot/