Re: [idn] iDNS re-chartering proposal, take 2

[Dave Crocker <dhc@dcrocker.net>]

At 02:42 PM 10/28/2001 +0900, Martin Duerst wrote:
>At 20:28 01/10/27 -0700, Dave Crocker wrote:
>>   the software doing the searching might not be a human, but it's place 
>> in the overall DNS system is the same place as human users.  That is, in 
>> the user agent, not the DNS infrastructure.
>
>User agent is a wide term, and may be okay. Human input/output
>is quite different, and inappropriate. As a human being I don't
>like to be thrown into the same pot as some software. So can
>we please change that to 'DNS input and output modules'?

I am not thrilled with the term "human i/o" either.  The difficulty is in 
using a term that clearly marks the DNS end-points (including DNS server 
administration, since that is where names are first entered.)

In the world of email and perhaps directories, the term "user agent" is 
pretty clear.  In the world of the DNS, it is not.

"DNS input/output modules" could as easily refer to communication between 
DNS servers, hence it is not sufficiently specific.

In case it is not obvious, I am at a bit of a loss, so the best I can do is 
suggest that we keep seeking suggestions for a term that will 
work.  (Frankly I would be delighted if we could find a way to re-apply the 
UA/MTA model, the way the x.500 directory people did.)

And that prompts me to ask whether simply saying:

         DNS Client

would satisfy people???

>I just wanted to make sure that the charter doesn't
>give the impression that all software has to use ACE throughout
>because conversion happens on 'Human input/output'.

Does "DNS client" satisfy this concern?


>>However the strings are created outside of DNS.  The only thing the DNS 
>>system must do is PRESERVE that order.
>
>DNS names appear on paper and displays. If we are not sure that
>the same internal string of characters will lead to the same display
>and vice versa, then all our work on idn will just lead to a lot of
>confusion. It may look as if just extending the set of characters
>won't change anything here, but at least for Arabic and Hebrew,
>that's not the case.

I fear that my point is being missed:

1.  The registrant decides what order to write the characters.
2.  That is how they are put into the DNS and returned from the DNS and 
displayed by DNS client software.

As long as the system preserves whatever order the registrant specified, 
then how does the DNS have to know whether it is arabic, hebrew, kanji, or 
greek?

In other words, I believe the DNS is, and must be, entirely opaque with 
respect to character ordering.

With respect to strings on paper or displays, again, the point is that it 
must be shown in whatever order the registrant specified, it.


>>>What's the difference between "Working Group action items are:" and "The 
>>>Action Item(s) for the Working Group are:"
>...
>I was asking something different, sorry. Why are there two subtitles in 
>your charter listing action items for the WG, one of them with two action 
>items, and one with four?

My turn to apologize.  I had not noticed that there were two sections with 
essentially the same labels.  Yes they should be changed, possibly 
eliminating one altogether.  I'll do something about that for the next 
version of the text.


>>>>Nov 2001        Draft of DNS character set enhancement specification
>
>Why not just call it 'DNS internationalization architecture specification'?

Your suggestion is for a term that is clean and impressive.  However I 
believe it also is too broad.

There are, perhaps, other issues in internationalization of the DNS that we 
are NOT working on.  Hence we need a term that excludes, as well as includes.

d/
----------

Dave Crocker  <mailto:dcrocker@brandenburg.com>
Brandenburg InternetWorking  <http://www.brandenburg.com>
tel +1.408.246.8253;  fax +1.408.273.6464