W3C home > Mailing lists > Public > www-talk@w3.org > January to February 2002

Re: discovering site metadata

From: Mark Nottingham <mnot@mnot.net>
Date: Thu, 7 Feb 2002 21:37:04 -0800
To: Dan Connolly <connolly@w3.org>
Cc: www-talk@w3.org
Message-ID: <20020207213657.A1858@mnot.net>

Well, I wrote up an I-D about this and everything, but after some
experimentation with Apache and IIS, it looks like supporting this
is... difficult.

Instead, what about just making a request to the most generic
resource on the server (i.e., /) and requesting the appropriate
content-type? So,

GET / HTTP/1.1
Host: www.example.org
Accept: application/p3p-prf+xml

GET / HTTP/1.1
Host: www.example.com
Accept: text/robots

406 Not Acceptable (or a response with a mismatching media type)
would indicate that it isn't available (in which case the legacy
well-known location could be polled).

(I chose OPTIONS * because that was an explicit request about the
*entire* server. After being lightly steeped in REST for a while, I
think that requests to / can make statements about all of the
subresources of / authoritatively. What I do want to avoid, however,
is having more than one entry point - it would be bad if you had to
request /images/ to see what metadata applied to subresources of it,
for example).

On Thu, Dec 06, 2001 at 11:48:32PM -0600, Dan Connolly wrote:
> On Wed, 2001-12-05 at 20:43, Mark Nottingham wrote:
> [...]
> > The obvious solution would be to use OPTIONS * along with an
> > appropriate accept header (or maybe something like AcceptNS?).
> er... not too obvious; I hadn't considered it before.
> It's a nifty idea.

Mark Nottingham
Received on Friday, 8 February 2002 02:51:14 UTC

This archive was generated by hypermail 2.3.1 : Tuesday, 6 January 2015 21:33:03 UTC