RE: Prelim. DAV spec.

It would be nice to just allow references to URIs as it simplifies the
BNFs. Furthermore the implementation of most servers, in so far as I am
aware, requires that each representation of a resource be individually
addressable even if they are accessable through a single URI.

However this does mean complicating the act of accessing a
representation. Instead of making a single request with the proper
headers to get to the desired representation we will either have to
provide an encoding of headers into URL space (probably through params)
or require that the user performs a HEAD and that the server must return
a content-location header. This sounds a bit messy.

I would like to hear some more discussion on this point before I alter
the spec.

				Yaron

>-----Original Message-----
>From:	Larry Masinter [SMTP:masinter@parc.xerox.com]
>Sent:	Thursday, October 31, 1996 9:57 AM
>To:	Yaron Goland
>Cc:	connolly@beach.w3.org; ejw@rome.ics.uci.edu; w3c-dist-auth@w3.org
>Subject:	Re: Prelim. DAV spec.
>
>Re my remark:
># When we add versioning, you can say that 
>#   a) versions apply to resources
># or
>#   b) versions apply to representations
>
># and there is some temptation, for generality, to try to go down the
># road (b), but I want to argue that choice (a) is the choice you should
># make.
>
>Yaron responded:
>
>> I say we take the weasel way out and support both. The draft already allows
>>for this.
>
>But actually, I meant to make a stronger case that you should NOT try
>to support choice (b): that is, don't try to allow for 'versioning' to
>apply to resources that don't have URLs. The reason that you shouldn't
>is that you'll have to invent some way of referring to the
>thing-that-is-versioned in order to talk about the version of it that
>you want, and that doing so will be hard, and you'll wind up with
>something that looks enough like a URL in the first place that you
>might as well have just required there to be a URL anyway.
>
>The primary reason why content negotiation ALLOWS for multiple
>representations without specific URLs for each of those
>representations is to support those situations where the
>representation is computed on-the-fly, e.g., different charset
>encodings (EUC, Shift-JIS) of the same Japanese document, different
>image representations (GIF, PNG, JPEG) of the same original image, or
>just different pages constructed on the fly from a database.
>
>I'd claim that in all of those situations, 'versioning' will not apply
>to the individual representations anyway.
>
>Are there any situations you can imagine or explain where the
>versioned representations don't conveniently have URLs but should be
>versioned independently?
>
>Larry
>

Received on Thursday, 31 October 1996 17:56:05 UTC