TODO: Machine-readable server announcements

"Daniel W. Connolly" <connolly@hal.com>
Errors-To: listmaster@www0.cern.ch
Date: Mon, 7 Mar 1994 19:02:13 --100
Message-id: <9403071733.AA04599@ulua.hal.com>
Errors-To: listmaster@www0.cern.ch
Reply-To: connolly@hal.com
Originator: www-talk@info.cern.ch
Sender: www-talk@www0.cern.ch
Precedence: bulk
From: "Daniel W. Connolly" <connolly@hal.com>
To: Multiple recipients of list <www-talk@www0.cern.ch>
Subject: TODO: Machine-readable server announcements
X-Listprocessor-Version: 6.0c -- ListProcessor by Anastasios Kotsikonas
Content-Length: 4525

This is a MIME message. If your software presents this
paragraph to you, it's not MIME compliant. See RFC1341 for details.

Well -- another explanation is that the www-talk list processor
seems to eath MIME-Version headers. Ack! Thtptptp!

--8<

There are a lot of mail messages and news articles of the
form

	"We're proud to announce a WWW server for
	avid bird-watchers at:
		http:www.bird-watchers.com
	"

I expect what follows is that lots of folks copy the URL
into their home page or their server's root page...

Isn't this an obvious candidate for automation?

This is another issue that the WAIS folks seem to have thought
about... their software comes with directions on how to send
a .src file to quake.think.com such that it automatically becomes
part of the directory-of-servers database.

Granted, the directory-of-servers becomes one huge flat database,
but WAIS is quite a good solution to that problem.

Anyway... I propose that folks who want to announce access to
an internet resource through mail or news use MIME to make
their announcement machine-readable. For example:


--8<
Content-Type: message/rfc822

To: www-announce@info.cern.ch
MIME-Version: 1.0
Content-Type: multipart/alternative; boundary="cut-here"

This is a MIME message. If your software presents this
paragraph to you, it's not MIME compliant. See RFC1341 for details.

--cut-here

We're pleased to announce our server...	
[blah blah blah in plain text]

--cut-here
Content-Type: text/x-html

<HTML><HEAD>
<TITLE>Bird Watching WWW Server Announcement</TITLE>
</HEAD>
<BODY>
<H1>Attention Internet Bird-Watchers!</H1>

We've collected lots of nifty info and made it availble on
<A HREF="http://www.bird-watching.com/">our server.</A>

Enjoy!

--cut-here--

--8<

This would allow folks to subscribe to www-announce (and/or
comp.infosystems.announce) and build knowbots to attack the
resource discovery problem.

We could even build a sort of _distributed_ database of
internet resources by having resource owners periodically
post announcements to comp.infosystems.directory-of-servers.

This could function as the "transponder" layer that
Weider and Deutsch dreamed up in
<ftp://ds.internic.net/internet-drafts/draft-ietf-iiir-vision-01.txt>

They drew a picture of internet publishing where
	* an author develops a resource
	* the author asks a publisher to publish the resource
	* the publisher allocats a URN from its namespace
	* the publisher installs a "transponder" that maps the URN
	  to some set of URLs (I guess this is a new TCP/IP service?)
	* folks can register interest in certain types of resources
	* the transponder notifies the folks of the new resource
	* those folks give the URN from the notification back to the
	  transponder in exchange for URLs
	* the folks access the resource

I think the vision can be realized using NNTP and local database mechanisms.
For example:

	* an author develops a resource
	* the author/publisher allocates a URN, for example, by generating a
	  message-id; something like: <x-message-id:1994.Feb.07lkjldj@foo.com>
	* the author/publisher installs a cron job that posts a
	  machine-readable announcement (be it an application/x-wais-source,
	  a text/x-html doc, an application/x-gopher-list, whatever)
	  to comp.infosystems.announce on a periodic basis
	* other sites install news-knobots that archive
	  comp.infosystems.announce (and process expiration... like FAQ
	  distribution) and provide local database searching
	  (e.g. a local WAIS database, or Fulcrum or Verity...)

The feature where users register interest in certain kinds of documents
(or in changes to certain documents, etc. ...) should be addressed in
a scalable fashion. A distributed notification system should be implemented
using a subscription-based strategy (look at Athena Zephyr and USENET NEWS
for successful examples). It's inefficient to maintain a mapping of each
of the N resources to each of the M consumers.

So, for example, we could establish a new branch of the USENET news
hierarchy for announcements about URN's (hmmm... or better yet, a whole
bunch of new leaves where appropriate; e.g. rather than
	urn.comp.sources for announcements about source archives, use
	comp.sources.urn )

The nifty part about this is that the question,
"What resources related to bird-watching have been
announced in the last week?" is already represetable in NNTP:
(I can't recall the exact syntax, but it's something like:)

	group rec.bird-watching.urn
	newnews 19940301000000Z


Comments?

Dan

--8<--