RE: Dereferencing HTTP URIs (redux?)

Hello Mark,

> I'm not entirely sure of the state of this discussion (for 
> which I apologise), but perusing 
> <http://www.w3.org/2001/tag/doc/httpRange-14/2007-05-31/HttpRange-14>
(which I understand 
> reflects, more or less, current consensus) leads me to be 
> somewhat concerned.

Firstly, you should, I think, understand that the document is a draft
and a work in progress and as such it's content does not necessarily
reflect a concensus of the TAG (let alone the wider community). The
status section does say "This document is an editors' copy that has no
official standing."

Such concensus as I believe exists is that *if* we (the TAG) are going
to discuss the document then, in the spirit of our imperative to work in
the public gaze, it is made publically visible. It is hard for us to
discuss something which is not.

My understanding of the TAG's intention is that it is *not* to unpick or
deprecate the advice that it gave when it resolved [1] httpRange-14. The
original motivation for the draft to which you refer was to provide
further elaboration of that advice in the form of a TAG finding.

[1] http://lists.w3.org/Archives/Public/www-tag/2005Jun/0039

> I've always been uncomfortable with the 303 solution that the 
> Semantic Web world has come up with for the "non-information 
> resource" problem. Inferring that two resources are related 
> in a fairly fundamental way because of a redirect between 
> them is IMO bad for two reasons;

Speaking personnally, I share some of your other concerns. However, I
think that, if so arranged, the use of 303 redirection is a useful way
to find additional information about a thing of interest. Of course in
the general case one cannot be sure that following a redirection will
yield a useful result. I would not expect to infer deeper relation
between the resources merely from the occurence of a 303 response. 

> 1) Re-defining the semantics of a core element in a protocol 
> that's been widely deployed for more than a decade will 
> surprise and displease some people.

On the whole I'd like to feel that where we end up would be a legitimate
use pattern for an existing facility.

> 2) The draft finding makes it a "good practice" to use 303, 
> when in fact metadata about the relationships between 
> resources may be available in much more efficient fashions. 
> For example, there's always site metadata, link headers, etc.
> 
> I'm sure this has been raised on www-tag before; I just 
> wanted to voice my concerns and then go hide under a rock again :)

:-) there are no rocks large enough ;-)

> 
> --
> Mark Nottingham       mnot@yahoo-inc.com
> 

Stuart
--
Hewlett-Packard Limited registered Office: Cain Road, Bracknell, Berks
RG12 1HN
Registered No: 690597 England

Received on Wednesday, 5 September 2007 13:11:14 UTC