- From: John Bradley <john.bradley@wingaa.com>
- Date: Fri, 12 Sep 2008 09:53:16 -0700
- To: "Williams, Stuart (HP Labs, Bristol)" <skw@hp.com>
- Cc: "Booth, David (HP Software - Boston)" <dbooth@hp.com>, "elharo@metalab.unc.edu" <elharo@metalab.unc.edu>, "www-tag@w3.org" <www-tag@w3.org>
- Message-Id: <D8643B4F-9C97-4C8B-90A9-3442077C866D@wingaa.com>
Hi Stuart, Thanks for the clarification. I think it is clear that a combination of Link header and 303 will be required to communicate that the URI is about a "thing" as David Booth puts it. I personally prefer to be a "thing" as opposed to a "non-information resource", or "abstract object". With your help we are slowly making progress towards at-least understanding each other. Regards John Bradley On 12-Sep-08, at 4:33 AM, Williams, Stuart (HP Labs, Bristol) wrote: > Hello John, > >> -----Original Message----- >> From: www-tag-request@w3.org [mailto:www-tag-request@w3.org] >> On Behalf Of John Bradley >> Sent: 11 September 2008 20:54 >> To: Booth, David (HP Software - Boston) >> Cc: elharo@metalab.unc.edu; www-tag@w3.org >> Subject: Re: [XRI] Back to XRI >> >> Thanks David, >> >> Certainly we want any http: subscheme to be completely >> backwards compatible with existing agents. >> >> In the case of a XRI like =jbradley where the XRI is a >> "identifier for an abstract object", me in this case. >> (Though thinking of myself as abstract still strikes me as a bit >> odd.) >> >> The question of what is returned when cast as a http: URI >> becomes interesting. >> >> In conversation with TAG members using Link headers to point >> at the URI for the XRDS meta-data seems reasonable. >> We currently use a 302 redirect to point at some appropriate >> html representation for compatibility reasons. In my case a >> contact page. >> >> Stuart Williams has pointed out that using a 303 redirect >> would make it clear that the resource >> http://xri.net/=jbradley itself is "an identifier for an >> abstract object" and not a resource that has temporarily moved. > > Not quite... you not conclude that it *is* an identifier for an > abstract > object, it may be but you cannot conclude that from the 303. > > What you can conclude (IMO), is: > > - that there is no claim (in the response) that the resource has > been moved > to a new location (that's different from 301/302) > - that you have not been returned a representation of the requested > resource > (for undisclosed reasons) > - that following the redirect (to something else) may yield relevant > information *about* whatever http://xri.net/=jbradley refers to. > - (probably) that the authority for http://xri.net/=jbradley in some > sense > warrants there redirection as useful/relevant to the original request. > > The Link header approach (with GET or HEAD) is a mechanism that > would work > similarly for both 'abstract' and 'concrete' resources in that it > provides a > mechanism where a metadata reference can be provided alongside a > representation in 200 response (indeed multiple references could be so > provided). > > [snip the next bit if you want to stayout of the philosopical tarpit.] > > In the venacular of the TAG and the httpRange perma-thread > 'abstract' and > 'concrete' correspond roughly to what have been called 'non- > information' and > 'information' resources (but even muttering that risks another turn > of the > wheel) - things that have mass (such as yourself) likely being > 'non-information' resources (or 'abstract' as you say - and as you > say that > seems odd). In fact in respect of the 'oddness' it would be > interesting to > get some idea of bounds on where you think the boundaries are between > abstractness and concreteness (though I suspect we will be no more > successful with that than distinguishing information resources from > non-information resources in a definitve way. > > From my pov... you are definitely *not* abstract, but there are many > facets > to you some of which maybe (eg. your life, or life history)- though > I might > wonder about say Shakespeare who has a similar concrete existence - no > longer does - 'my notion of Shakespeare' is certainly abstract - but > if you > and I are referring to Shakespeare are we referring to a concrete > thing or > an abstract thing or... might we just not know or in many cases not > care to > make a distinction? > > [If you really want to start making your head hurt there's an > interesting > dialog about 3d/4d conceptualisation of the world recorded at [1] > and then > its probably off to metaphysics 101... and probably a very long > pause...]. > > [1] http://www.adampease.org/Articulate/dialog-3d-4d.html > >> The XRI TC originally selected the 302 redirect for >> compatibility with pre http 1.1 browsers. >> Changing to 303 redirects may brake some clients but it >> unlikely to be a significant issue. >> >> I understand that some people will feel that using redirects >> is inefficient, however it seems the only way to communicate >> the desired qualities of the identifier in http: > > What 303 does as opposed to 301/302 is make no assertion of > relocation of > the reference resource. It simply says "See other". > > Whilst on the one hand that is a very weak claim, bearly more than a > hint > that there *may* be useful related information elsewhere, it can be > used to > useful effect. > > HTTP Link: headers with a more distinguished rel value could be used > to make > a stronger claim (by virtue of what is entailed in the specification > of a > given rel value eg. rel=http://xri.net/relations/xrds-metatdata > could be > specified to be an assertion by the relevant (URI) authority that > related > XRDS information *is* or at least *should be* available at the link > target. > >> This still allows XRI aware applications from directly >> requesting the XRDS meta-data if they recognize the subscheme. >> >> I will also mention that some people seem to prefer the term >> "http: profile" as a way to describe the qualities of a sub scheme. > > Personnally, I just see it as a way that some authority architects > the use > of the URI subspace under its authority - be that *.xri.net or *.xri > (were > that TLD to be sought). > > BTW: I'm sure that I spotted a URI of the form something like > library.hpl.hp.com.<service>.<otherdomain>.com which seem to suggest a > similar pattern of use. Wish I'd been more careful to not thee > particular > URI at the time. > >> Regards >> John Bradley > > -- > Hewlett-Packard Limited registered Office: Cain Road, Bracknell, > Berks RG12 > 1HN > Registered No: 690597 England
Attachments
- application/pkcs7-signature attachment: smime.p7s
Received on Friday, 12 September 2008 16:53:57 UTC