Re: [idn] draft-klensin-dns-role-00.txt

["Martin J. Duerst" <duerst@w3.org>]

Hello John,

I have read the paper and the discussions recently,
and I have a few comments.

First, I think the paper is very well written and
compiles ideas, thoughts and opinions that have been
around for quite a while in a way I haven't seen yet.

However, I also have some questions.

The main confusion that I see with the current draft is
that it implicitly distinguishes search, directories, and
lookup, but especially the directory area isn't very clear
for me. It seems that the assumption is that a directory
will return more than one result per query. In some
cases, that's helpful, but the assumptions in most cases
of using the DNS are different. People don't want to
select from a list each time they send out a mail,
and they don't want to select from a list each time
they follow a link on a Web page.


Some more general comments: Nobody really thinks that the
DNS is perfect, and we all wish we could redesign it,
and every time somebody comes up with something that may
need fixing (like i18n now), people start dreaming and
discussing about such a redesign, and try to see whether
the fix under discussion might be the chance for the
redesign. Usually, it's not, because the fix is needed
with some urgency, the people needing the fix don't
want to do the redesign work, and there are too many
things that are unclear (e.g. will the redesign be
better? will it ever take off?). I have been trying to
think about examples where internationalization was used
as a lever for ground-up redesign, but I haven't found it
yet. I'm not sure whether that's due to the general
problem of redesigns or is to some extent specific to
internationalization.

Also, once something is built, it starts to be (mis)used
for what it is good at, even if it was not designed for
that. In some sense, the original design goals become
secondary, even if it's always very interesting to
understand them. For example, if the flat namespace
(in .com,...) hadn't scaled that well (which is at least
as much due to hardware improvements than to DNS itself),
DNS just wouldn't have been used that way.

One thing the draft proposes is to work towards some kind
of directory. I know (just from hearsay, I have to admit)
that there are various directory solutions around, but none
would really just be ready to be used for DNS, and none of
them is as ubiquous as e.g. DNS, email, and the Web. It would
be very good if somebody with actual experience in that field
would write a document discussing the reasons for why this
is so, and what we might learn from it. (or send a pointer
if such a doc already exists)


Regards,   Martin.


At 00/12/18 05:07 -0500, John C Klensin wrote:
>--On Tuesday, 12 December, 2000 10:15 -0800 Hideyo Imazu
><hideyo.imazu@msdw.com> wrote:
>
> > Hi.
> >
> > I observed John Klensin's presentation at the Monday slot of
> > IDN in the  IETF meeting.  I also read through the IETF Draft.
> >...
> > Having said that, my interpretation of his presentation is
> > "now is the  time to rethink and overhaul the current resource
> > locating schemes in  which DNS has the central role."  After
> > the overhaul we might end up  having a next generation DNS
> > with directory like feature or some  hierarchical directory
> > scheme with an innovative design concept.  Or we  might
> > introduce a generic resource locating layer between
> > application  and DNS.  All Internet applications including web
> > and email would  eventually be based on the new resource
> > locating scheme.
> >
> > John, is this what you really mean?  Do you have any
> > particular plan to  form a BOF or WG for it?  What other IDN
> > folks think?
>
>That is, indeed, what I really meant.  Still exploring plans
>--with the IESG and others-- about how to pursue this.  I'd
>personally be happy to have the IDN group take it up, but I
>infer from Wednesday's meeting that it will need to be done
>elsewhere if at all.
>
>Suggestions welcome.
>
>     john