W3C home > Mailing lists > Public > www-zig@w3.org > February 2002

ExPLain + Friends & Neighbours -- propose a working group

From: Sebastian Hammer <quinn@indexdata.dk>
Date: Mon, 25 Feb 2002 11:09:09 +0100
Message-Id: <4.2.0.58.20020225104241.01a4c998@bagel.indexdata.dk>
To: <www-zig@w3.org>
Hi,

Thought it was time to drop that misleading subject line.

It seems to *me* that the consensus that is appearing is this: There is a 
requirement for a basic way to describe targets, their network addresses, 
search capabilities (etc.), and a bit of content information as well. We 
could see this information distributed using an Explain-like search 
mechanism, through an Init-exchange, through HTTP from the server host, or 
through some other mechanism from third party target lists. As Eliot points 
out, it could also be made available through a mechanism like UDDI, given 
an appropriate schema mapping. It seems to me that there'd be good reasons 
why roughly the same data model should be useful in all cases. Even more, 
some people have suggested that maybe the Friends & Neighbours model still 
deserves a shot... if that, too, can be thought into the same model, than 
it would be a "free" service.

To me it seems like there's good reasons to design the data model first and 
then map it onto exchange mechanisms later. Specifically, I would say it 
would be putting the cart in front of the horse to worry about the 
UDDI/WSDL representation of searchable fields before we have agreed what it 
actually is we want to represent.

It would seem to me essential that the model be flexible, both in terms of 
extensibility and minimalism. All that is required for a Friends & 
Neighbours service is a list of host addresses. More ambitious, a 
semi-automatic third party server might have addition information about 
things like searchable fields and record syntaxes, but it would not 
necessarily have much content-descriptive information, such as might be 
supplied by a server's own "explain" function. In the other extreme, we 
probably want a model that can be easily extended as well for local purposes.

Could I ask those who agree with this general approach, and who would be 
interested in shaping the model to drop me a line, off list? If there are 
any takers, we can knock up a little mailing list and see if we can carve 
something out in time for discussion at the next ZIG. Mind you, I don't 
propose to lead the discussion or seek credit for the result -- just to 
kick things off, and nudge knowledgeable people to provide input, if 
necessary. Some with previous explain (lite) experience would be welcome 
(but such experience is obviously not a requirement!). Similarly, it sure 
would be nice to have someone along with some level of good XML design 
expertise.

--Sebastian
--
Sebastian Hammer, Index Data <http://www.indexdata.dk/>
Ph: +45 3341 0100, Fax: +45 3341 0101
Received on Monday, 25 February 2002 05:08:53 UTC

This archive was generated by hypermail 2.3.1 : Wednesday, 5 February 2014 07:13:27 UTC