18 Mar 2008

See also: IRC log


+1.781.643.aaaa, dbooth, Alan, Jonathan, stuart, TimBL




Properties that relate an HTTP response to the original URI

<jar286> http://www.w3.org/Protocols/rfc2616/rfc2616-sec6.html#sec6

Jonathan: What I'd like to do today: A property that relates an HTTP response to the thing denoted by the original URI that you were asking about.


jar: when you do a get, you should learn something about the URI.

<jar286> about the thing denoted by the URI

alan: Assuming no conneg, and a fixed MS word doc is sent. how do we know how to interpret it? Need some metadata to know it's a word doc. (ignoring http spec for the moment). What's a sufficient knowledge rep of that? A binary blob with a typing triple.

<alanr> "alksdkakd;ad" rdf:type mw:word_document

<alanr> is this the microsoft word document information resource?

?blob rdf:type mw:word_document

<alanr> q: let's define one simple informationr resource, independent of http

<jar286> ?blob foo:has_content "0100100100100101"

http://example#foo :hasRepresentation ?blob^^mw:microsoft_word

<alanr> "http://example#foo" :denotes

oops, no, because the ?blob itself is supposed to have the data type.

<jar286> jar gets alarmed whenever we talk about "information resources"

<jar286> how about information thing

<jar286> for the purposes of this conversation

Alan: http://example/mydoc denotes a thing that consists of a type and a blob.
... or a typed blob.
... Now we can talk about what happens with HTTP.

<alanr> http://office.microsoft.com/en-us/ork2000/HA011382051033.aspx

<alanr> there is a mime type called application/msword

<alanr> get http://example#foo ->

<alanr> response has a number of headers

Alan: Should be GET http://example/mydoc

<alanr> mything = that ms word document information thing

<jar286> (danger: I would predict Tim to reject http://example/mydoc leading to a 200 response for the same reason that he rejects a number leading to a 200 response - it is a mathematical object, not a Fieldingesque 'resource')

<alanr> my goal is to get a message that lets me make an exact copy

<alanr> hi and bye stuart

<alanr> please call in

what is the relation between a http response and the thing it denotes

<alanr> it = the URI

<Stuart> 'origninal'?

Alan: I have a fixed MS word doc. It's a basically a typed blob.
... Want to be able to send an exact copy of that thing over HTTP.
... So you would GET from a URI, and you'd get back a mime type and an entity body.

<alanr> http://example#foo

Alan: I can use the entity body and the mme type to reconstitute the typed blob.

no, to start simple, use this uri: http://example/mydoc

(Because you can't GET from a UIRI with a frag ID.)

So if the document owner puts the document at /mydoc on server example, then GET http://example/mydoc will return what you need.

timbl: In the WebArch, if the MS Word doc is the vitals of the society, the way the web's used, you say the URI identifies the vitals of the society. You don't say it identifies a word doc. What's interesting is that it's the vitals of the socieity. Then a GET response is a representation.

<Stuart> I think that the response 'convey's a representation of the resource.

<jar286> timbl (jar's paraphrase): maybe a word document is not a good example because we don't use uris to name the kind of thing this discussion started with

Alan: I was trying to make life simpler -- assuming no conneg.

<Stuart> exists resource R such that response RESP conveys representation REP such that REP represents (stateOf) R (at time of request)

Alan: So I notice if I do a GET on http://example/mydoc the entity body and mime type would allow me to reconstitute my object.

<Stuart> reconsitute == create a clone?

Alan: Then we could say that in that case, the relationship of the thing denoted by the URI and the response you get is this: you take the entity body and the mime type, and you interpret them as the blob and type.
... What other reasonable responses might there be?

<Stuart> Alan: The final phrase of what you said tailed of in "that thing" and I don't know whether that is a reference to the original document or its clone - indexicals!

Alan: The server could return an HTML doc instead.

<jar286> "that thing" = the information thing = particular bits + type information

timbl: no, if the doc is a fixed resource, then you cannot give something else back.

<jar286> stopping for homework in 2 minutes

<Stuart> jar: at best, in webarch, bits+type info are a representation (of ??) [the (current?) state of the resource]/

<Stuart> at best I see that as a type for a set of self similar messages.

<jar286> stuart: but this discussion didn't start with uris, or representations, or resources. it started with this information thing

Alan, now supose the server gives a 303 response instead of 200.

timbl: why do you want a 303?

Alan: I dont' yet know it is a fixed resource, and I don't know yet it is a set of bylaws.

<timbl> Link:

Alan: So how would I get that other information?

jar: If each individual in the group could produce an account (in RDF preferably) of relating a GET response to the thing denoted by a URI, we could compare them.

Alan: yes, but there are different kinds of HTTP responses. So if it wasn't a fixed resource, there are several things you could get back.

jar: Write as many relationships as you can think of.

<Stuart> Alan, can you indicate what conclusion you are trying to reach...

dbooth: My rules show the relationship between the thing denoted by a URI and a URI declaration, which sounds very much like what you are asking for.

jar: Yes, but i want to see it for documents.

dbooth: My rules did not cover the document case.

<Stuart> are we talking about a form of typed literal?

<Stuart> [i don't mean an RDF typed literal or an XML literal]

stuart, you can call them something else if you want (not UIR declaration), but it is a collection of metadata about the denoted resource.

<jar286> dbooth, sorry to hijack your agendum.

no prob.

<jar286> the discussion was *inspired* by your work...

it's amazing how we're still struggling so much with such basic things.

<jar286> yes

very hard to get a common starting point.

<jar286> are you going to close up with rrsagent?

jar, yes i'll post the minutes, incomprehensible as they may be :(

Summary of Action Items

[End of minutes]

Minutes formatted by David Booth's scribe.perl version 1.133 (CVS log)
$Date: 2008/03/18 15:39:35 $

Scribe.perl diagnostic output

[Delete this section before finalizing the minutes.]
This is scribe.perl Revision: 1.133  of Date: 2008/01/18 18:48:51  
Check for newer version at http://dev.w3.org/cvsweb/~checkout~/2002/scribe/

Guessing input format: RRSAgent_Text_Format (score 1.00)

Succeeded: i/Meeting/Topic: Properties that relate an HTTP response to the original URI
Succeeded: s/get/give/
No ScribeNick specified.  Guessing ScribeNick: dbooth
Inferring Scribes: dbooth
Default Present: +1.781.643.aaaa, dbooth, Alan, Jonathan, stuart, TimBL
Present: +1.781.643.aaaa dbooth Alan Jonathan stuart TimBL
Regrets: Noah
Got date from IRC log name: 18 Mar 2008
Guessing minutes URL: http://www.w3.org/2008/03/18-awwsw-minutes.html
People with action items: 

[End of scribe.perl diagnostic output]