<jar> i started a page http://esw.w3.org/topic/AwwswAnalysis of first principles analysis... consensus i hope

<jar> http://esw.w3.org/topic/AwwswTopicsBrainstormPage

<jar> Crude beginning on a consensus document http://esw.w3.org/topic/AwwswAnalysis

<scribe> scribe:alanr

tim: can't define 200 exactly, in english - would be a mistake. Take a tutorial. Define in terms of how interactions go in protocols
... Defining what information resource is, is a trap
... What's useful is the content of the message (?)

what happens when you get a 302 then a xxx and then a xxx, what does that mean?

<dbooth> I like the idea of being able to explain what it means when you get a 303 followed by a 302 followed by a 200.

tim: tabulator does generate triples and does use them

<timbl> Tim: What is useful is the semantics of the message-- eg what a 200 means in terms of what [how] the body [is] related to the abstract document.

jonathan: we should address tabulator's needs, but generalize

and what's a document?

Jonathan: didn't see what 200 concluded

<Stuart> methinks.. something with a document oriented interface :-)

Tim: concludes it's a document, and opens up a document view. For image/ opens an image view

tim 303: won't give you an option of looking at the document

tim: 302 generates, as part of document containing the link, gives warning

<Stuart> eg. a [remotely controlled] robot arm could be controlled via an exchange of 'document-like' messages.

<Zakim> dbooth, you wanted to say that key to me is putting proposed RDF on the table of what these things mean

<Stuart> can the URI 'identify' a robot arm rather than a document 'implied' [by] the message exchange.

<timbl> { ?x http:resresentation ?r } ..

jonathan: Are the tabulator's rules in one place?

<timbl> http://dig.csail.mit.edu/2005/ajar/ajaw/js/rdf/sources.js

<timbl> FWIW the code in which tabulator has hard coded inference

<timbl> if (xhr.status-0 == 200) {

<timbl> addType(tabulator.ns.link('Document'));

noah: asking about image/

<dbooth> noah: [suppose] I have a resource that I know is a text doc, but for whatever reason I have a photographed copy of the page. This seems to imply that the media type bounds the nature of the document to the image.

<dbooth> timbl: Yes, that's not quite right, but it's a good hack.

<timbl> Note addType pecolates back up through the 301 an 302 and adds thr typre s to them too

<timbl> The conclusion that something 'is' an image from the image/* content-type is a hack, not architecturally solid.

<dbooth> noah: We should say that the nature of the media type does not further bound the nature of the resource.

<timbl> link:Document owl:disjointWith foaf:Person.

<Zakim> noah, you wanted to ask tim about 200

<jar> stuart: robot arm responding with 200s reporting position, is the robot arm a document? (tabulator says yes)

<dbooth> "What do you want to say about it?" is very much like "what inferences do we want to make?"

<jar> noah: this is just httprange-14

<Zakim> alan, you wanted to ask: httpRange-14 says "If an "http" resource responds to a GET request" - does this tell us what a resource is - it can "respond"

<dbooth> My opinion: To be kosher, the robot URI that returns 200 would be the URI of the HTTP *interface* of the robot.

<jar> jar: let's talk about 200 where we already believe the thing is a document, not other cases, for now

is a form that lets you tweak a robot, a document?

<dbooth> I agree with starting with "What inferences do we want to make?", and I want to see RDF on the table.

http://www.w3.org/2001/tag/issues.html#httpRange-14

<jar> alanr: httpRange-14 is unclear - what was the intent re the language? is it really the resource that responds?

agreed on 15 Jun 2005

The TAG provides advice to the community that they may mint "http" URIs for any resource provided that they follow this simple rule for the sake of removing ambiguity:

* If an "http" resource responds to a GET request with a 2xx response, then the resource identified by that URI is an information resource;

* If an "http" resource responds to a GET request with a 303 (See Other) response, then the resource identified by that URI could be any resource;

* If an "http" resource responds to a GET request with a 4xx (error) response, then the nature of the resource is unknown.

<jar> timbl: it's not cleanly put. should say ... something about the server ...

<jar> alanr: requests some action [on the part of the TAG] on this language

<dbooth> Should be something like "If the HTTP status code resulting from a GET request on a URI is 2xx . . . "

<jar> timbl: ok [that the TAG should act to clarify the language]

<Stuart> FWIW... I prefer to think of HTTP requests as questions being asked of Web Infrastructure and responses as being generated by web infrastructure.

<jar> http://esw.w3.org/topic/AwwswHome

<Stuart> not all requests make it to the web server (caches may respond instead);

<dbooth> I want our main work product to be RDF and inference rules.

<dbooth> I think other wording improvements may fall out from that.

<dbooth> I have put RDF on the table, and will continue to do so. I haven't looked specifically at what tabulator does.

http://esw.w3.org/topic/ErrataHttpRange14

<jar> jar: asking timbl what best process is for getting tabulator's inference rules recorded in the wiki. timbl short on time

<jar> timbl: is putting them in now

<timbl> http://esw.w3.org/topic/AwwswAnalysis

<dbooth> jar: Some agents may want to be conservative in the inferences that they make, in case the server is hostile, while others may be more liberal, like tabulator.

<jar> alanr: jar is talking about default reasoning, perhaps. aggressive inferences are defaults

<dbooth> alanr: You can have a logic in which the aggressive things are assumptions based on getting extra info.

<jar> Discussion of boundary cases is a quagmire. Example of an httpRange-14 boundary case: the number three. not a boundary case: a person.

<jar> alanr: wants to be clear: are you manipulating the form or manipulating the robot? ambiguous -

<jar> timbl: a form is just like a document

just dropped. will call back

yes

<jar> noah, i can't do justice to what you're saying

<jar> tim wants to say "document" instead of "information resource", noah doesn't

<Zakim> timbl, you wanted to suggest some more uses for having the classes allocated to things.

tim: having disjoint classes gives teeth to this [effort]

<jar> timbl: tabulator is becoming an editor - disjoint classes will matter

tim: tabulator is an editor now. The more you know about this the thing the better the editor can help you

<dbooth> i have to drop off the phone now, but i'm strongly in favor of: 1. specific use cases; and 2. Discussing some specific RDF / rules. I think these will best help us forward.

+1 to warnings

<jar> stuart: can we do some backward chaining - start with inferences one would *like* to be able to do

<dbooth> I would like us to discuss the RDF / rules I've drafted next meeting.

<jar> timbl: alan threw down the gauntlet - how do i know whether a web server has lied?

<dbooth> Though I may have a better version by then. :)

<jar> stuart doesn't believe this is a small enough question

<jar> alanr: e.g. "this resource's responses will always be images" - a non-image response would be wrong

<timbl> Narrow escape from the Modelling Time rathole, a fine hole of great, and not uninteresting to explore.

so, combination of statements about a resource, including, but not limited to http responses, should be able to say something wrong