Re: [idn] New protocol proposal: IDNRA

[Keith Moore <moore@cs.utk.edu>]

>   While I don't have any strong evidence to contradict the pessimistic
> rates of comformant adoption that Keith and John (and others) are
> predicting, I have to wonder if things in the world of Internet computing
> have changed (or are changing).  That is to say that the pessimistic view
> is based largely on historical experience with SMTP, MIME, etc. and
> perhaps some of the factors that created that set of problems aren't true,
> or as true, for this area.

I don't see it as pessimistic.  I see it as optimistic, because we really
do have a chance to get IDNs finished and into products before a lot of 
the non-English speaking world gets on the Internet.  The caveat is that 
we need to beat that deadline; otherwise, it will be much more difficult 
to deploy IDNs.

but we do need to realize that there are factors which make IDNs
difficult to deploy (e.g. difficulty of replacing installed base), 
and other factors which aid their deployment (e.g. if we can get
them done before most of the non-English speaking world gets on the
Internet).  it's not all gloom and doom; it's a mixed situation.
 
>   I wonder if given some of the relatively newer pressures in the domain
> of Internet technology adoption, namely increased commercialization (would
> that be c15n? :-), an increasingly global focus. the i18n issues we've
> been discussing, demands for higher levels of QoS, and very high quality
> open source software, that a rollout of a multilingual DNS protocol might
> be a less dismal prospect.

I think MIME is a fairly good indicator.  MIME was designed to be deployable
at the edges while leaving the core alone; thus the deployment was in the
hands of those who will benefit. presumably IDNs will have a similar model.  
Although the early MIME implementations were open source, they were (and 
frequently still are) quite crude in terms of their user interfaces; 
most of the drive behind the deployment of high-quality MIME clients was 
commercial in nature.  (MIME email was also able to benefit from the WWW's 
popularity because they used a common typing system and because many web 
browsers were also designed to be mail readers.)

note that the problem isn't implementation so much as deployment.
it is probably fairly simple to add support to display and input 
of an IDN to a MIME mail client that already has I18N support.
but it is often quite difficult to get users to change mail clients,
or sometimes, even to upgrade.

Keith