W3C home > Mailing lists > Public > public-awwsw@w3.org > July 2009

Re: FW: Http.ow and jar-diagram-7.pdf l (was RE: JAR conflict for July 7 AWWSW telecon)

From: Jonathan Rees <jar@creativecommons.org>
Date: Mon, 6 Jul 2009 11:14:10 -0400
Message-ID: <760bcb2a0907060814m330e6897k5d58a0921a738b94@mail.gmail.com>
To: "Williams, Stuart (HP Labs, Bristol)" <skw@hp.com>
Cc: AWWSW TF <public-awwsw@w3.org>
On Mon, Jul 6, 2009 at 9:55 AM, Williams, Stuart (HP Labs,
Bristol)<skw@hp.com> wrote:
> [Apologies accidentally hit the send button before...]
>
> Hello Jonathan,
>
>> I would still really like to get feedback on
>> http://w3.org/2001/tag/awwsw/http.owl
>> and
>> http://w3.org/2001/tag/awwsw/jar-diagram-7.pdf
>
> The one relation that I can't make sense of (yet) in http.owl is "http:ofWaRepresentation" the associated comment says:
>
>        "C hasWaRepresentation R when R is the wa-representation that corresponds to the resource in which C inheres during C's lifetime."
>
> which seems to be a comment about a different relation... Or at least has not kept up with some name changes. Domain and range of "ofWaRepresentation" are "http:Message" and "http:Entity" respectively which has left me confused about the intent of the relation.
>
> Jar-diagram-7.pdf shows the same relation as having a domain of "Correspondence" and a range of WaRepresentation. This makes way more sense than it does in http.owl... Ie. There is a correspondence between the WaRepresentation and a Rfc2616Resource (Get200Candidate) evidenced by a message exchange which conveys the WaRepresentation.

Fixed I think.

> --
>
> Re rdfs:comment on http://w3.org/2001/tag/awwsw/http#WaRepresentation ie.
>
> <quote>
> This is meant to capture the web architecture definition of 'representation', which is a problem because there isn't complete agreement on what this is.
>
> AWWW has "Data that encodes information about resource state" but this does not necessarily mesh well with HTTP, which does not require either that the resource has state or that a 200 entity encodes information about state when it exists.
>
> HTTP has "An entity included with a response that is subject to content negotiation, as described in section 12. There may exist multiple representations associated with a particular response status."  We also find (at least in HTTPbis) "Any response containing an entity-body MAY be subject to negotiation, including error responses." I guess it's up to the server, and the entity/representation distinction is not operational.
>
> Some people (e.g. Harry) have suggested that representations *cannot* be persistent (stored) - that they are "on the wire" only. I don't see this being supported by either definition.
>
> I'm hoping we'll be able to come up with a consensus definition that bends each of these enough to harmonize them - or else that we decide we don't need this class at all for a treatment of HTTP.
> </quote>
>
> I think this is a comment about a type/token distinction for WaRepresentations and when two distinct messages convey the same representation. One view is that they simply can't because two distinct messages must occupy a different region of space/time. The other view is that as long as they convey bytewise identical content they convey the same representation. Somewhere in between, somewhat more coarse grained than the first of the position above and more fine grained than the second is to identify a representation by it's bit/byte sequence (inc media type) and last-modified time/date. This is then reflective (or should be) of when the server first emitted the representation in response to a question - and respects the repetion induced either by multple agents asking (effecitively) the same question and being served the 'same' answer be it directly from the origin server or a cached copy from a proxy.
>
> [However, beware transcoding proxies and local ad insertions - which may change representions without reference to their origin server]
>
> Summary: Suggest WaRepresentation identified by (bytesequence)x(media-type)x(last-modifed date/time) - there's a role for OWL 2 compound keys :-)

Including last-modified but not expires seems oddly asymmetrical to
me. I would rather remove the date/time so that we can more often say
the same representation corresponds to multiple resources, a la the
genont theory. I have no idea how carefully last-modified is
maintained by servers and proxies; perhaps what you suggest will work.
Let's try to figure this out over the coming months...

I also include Content-Language in the identity of a representation.
See new comments under HttpRepresentation.

> --
>
> http:Entity
>
> This a strange one... http:Entity are encapuslated in http:Messages ie. They are a *partOf* an http:Message. What they carry may be *partOf* a WaRepresentation (amongst other things). In that respect I can't really see an http:Entity as rdf:subClassOf http:WaRepresentation.  [I may be confusuing message body with entity - I'm thinking of chunked situations where an http:Message conveys less than a full WaRepresentation].
>
You're right, I've flushed this idea, thanks. Have introduced
HttpRepresentation as AKO WaRepresentation, hoping that this works.

> --
>
> Jar-diagram-07.pdf includes a (tentative?) notion of a process. I think that process may serve to surface 'evidence' of correspondences in the form of http interactions, but with a god-like eye such correspondences exist whether or not the action of a client cause them to be exposed. Hmmmm... but what about the actions of a published that publishes the resource - without their actions the correspondence certainly could not be observed - and may not exists at all. Hmmmm....

The process is meant to be something involved in the existence or
maintenance of the resource - e.g. the weather in Oaxaca, or the
process leading to the recent transactions list in your bank account
view, or a process inside of W3C that produces successive drafts of a
document.

> Maybe the "Exchange" in jar-diagram-7.pdf is induced by the process shown therein (the process being part of the behaviour of a "Client"?); likewise for http.owl although process is currently absent from the file.

GET is side-effect-free ideally, so the process around fetching and
interpreting representations has to be completely independent of the
process that leads to correspondences. Both processes, yes, but
different ones.

> --
>
> In http.owl:
>
> I find the little clutch of http:Get200Candidate subClasses (transitively) confusuing - I'm a little lost about the intent of each and the distinctions between them

The name and comment for GoodGet200Responder was wrong ... needs to
mention httpRange-14. I made a few more tweaks that might help. I
admit it's a mess, but it isn't my fault - I'm just trying to be a
faithful recorder of what has been proposed.

> HTH

It does! thanks for listening.
Jonathan

> Stuart
> --
>
>
>> -----Original Message-----
>> From: Jonathan Rees [mailto:jar@creativecommons.org]
>> Sent: 06 July 2009 12:12
>> To: Williams, Stuart (HP Labs, Bristol)
>> Cc: AWWSW TF
>> Subject: Re: JAR conflict for July 7 AWWSW telecon
>>
>> On Mon, Jul 6, 2009 at 5:13 AM, Williams, Stuart (HP Labs,
>> Bristol)<skw@hp.com> wrote:
>> > I too have a conflict and cannot make the call on 7th July.
>>
>> OK, it looks as if 7 July is canceled.
>>
>> I would still really like to get feedback on
>> http://w3.org/2001/tag/awwsw/http.owl
>> and
>> http://w3.org/2001/tag/awwsw/jar-diagram-7.pdf
>>
>> but I don't know if it's worth scheduling an extra call. Let's try
>> email, and I'll meet with Alan R in person about it.
>>
>> I want to start thinking soon about what should be in a report on this
>> effort. Not that the work is anywhere near done, but I hope each of
>> you (active participants) will think about what would be good content
>> and structure, to help guide next steps.
>>
>> Jonathan
>
Received on Monday, 6 July 2009 15:14:51 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Monday, 6 July 2009 15:14:51 GMT