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

RE: Are generic resources intentional?

From: Williams, Stuart (HP Labs, Bristol) <skw@hp.com>
Date: Thu, 11 Jun 2009 11:02:01 +0000
To: Pat Hayes <phayes@ihmc.us>
CC: David Booth <david@dbooth.org>, Alan Ruttenberg <alanruttenberg@gmail.com>, Jonathan Rees <jar@creativecommons.org>, "noah_mendelsohn@us.ibm.com" <noah_mendelsohn@us.ibm.com>, AWWSW TF <public-awwsw@w3.org>
Message-ID: <233101CD2D78D64E8C6691E90030E5C832D1C344FC@GVW1120EXC.americas.hpqcorp.net>


> -----Original Message-----
> From: Pat Hayes [mailto:phayes@ihmc.us]
> Sent: 10 June 2009 16:40
> To: Williams, Stuart (HP Labs, Bristol)
> Cc: David Booth; Alan Ruttenberg; Jonathan Rees;
> noah_mendelsohn@us.ibm.com; AWWSW TF
> Subject: Re: Are generic resources intentional?
>
>
> On Jun 10, 2009, at 9:50 AM, Williams, Stuart (HP Labs,
> Bristol) wrote:
>
> > Opps... a vital missing speech mark... and an acknowledgement that
> > this is *my* rough framing of what I take Alan to be asking...
> > Rather than an assertion that this is exactly what Alan asked or
> > said. Apologies, particularly to Alan, if that was mis-understood.
> >
> >>> So... we're back roughly at the nub of Alan's question, roughly
> >>> "...which is it, there resource of which we speak... the
> >>> (passive)document/work of which the server wa-representations are
> >>> representations of; or the (active) agent'y entity that provides
> >>> responses to questions." I believe that the traditional
> view is that
> >                           ^
> >>> the URI names/identifies/refers-to the document/work thing rather
> >>> than the (conceptual) machinery in the web (which some have dubbed
> >>> http-endpoint).
> >>
> >> Right, sorry I missed that earlier question. Good
> question, but Alan
> >> gives exactly the wrong answer. I'm not sure what tradition he is
> >> referring to, but the REST/tag/awww answer is surely that
> it has to
> >> be
> >> the active agent'y entity.
> >
> > It is *I* not Alan
>
> Apologies for getting it wrong, but really Im arguing against the idea
> here rather than its author :-)

Indeed... I was just not wanting to have appeared to have put words into Alan's mouth.

>
> > making a reference to a 'traditional' view and what I'm referring to
> > really just the notion that (http) URI (are/can be used to) name
> > documents which pervades most of the writing I seen on web
> > architecture. To be fair, Roy does/has taken the position that there
> > are more things than documents on the web, ie. active, agent'y,
> > entities with which one interacts through the exchange of wa-
> > representations (which might convey wa-representations of
> > current(get) and desired(put/post) state - though I've seen
> no such
> > constraint spelled out).
>
> And he is right, of course. And I think there is an intellectual
> imperative at work here, which I respect, to avoid 'limiting' the
> categories so as to exclude possible future technologies. I recall Roy
> talking about the view of a city from a webcam meaning that the city
> was in a sense 'web-accessible'.
>
> Here's my problem/argument with the idea that its all about the
> documents. Suppose I have a computer which  isnt connected, or even
> connectable, to the internet. I create a document - say, just to be
> bloody-minded, an HTML document - on it. The document/file/whatever
> exists, no doubt about that. It has all the properties it will ever
> have **by virtue of being a document**. All its 'essential'
> properties, the ones that inhere in it by virtue of its being the kind
> of thing that it is, are present in it now. I can even give it a name,
> and - again to be bloody-minded - choose a name which has the syntax
> of a legal URI which I happen to own. And I can perform if you like a
> public ceremony, witnessed by others and with a video posted on Utube,
> naming it, a document baptism. None of this actually gets it onto the
> Web, though.

Ok...

> Ok, finally I relent, and put this document on a server
> and make it be the payload of a GET request with that URI. Ah, **now**
> it is on the Web. But, and this is my point, nothing **in the
> document** has changed here. It is the **very same document** that is
> was before I put it on the server, and it even has the **very same
> name** that it had then. So whatever it was that 'put it' onto the Web
> cannot possibly have anything to do with (a) its nature as a document
> nor (b) the name I used to refer to it with. Because these have not
> changed, but its Web status has changed. So whatever that status
> inheres in, it must be something else, something other than the
> document/name pair itself. And it seems, just being intuitive now,
> pretty obvious that it inheres in the document being **on a server**.
> Its the presence of the server that makes it into a webbish object.

Ok... you've reminded of a message from Joshua Allen early in the life of the TAG which introduces the notion of resources as (wa-)representation o hypertext *dispensers* http://lists.w3.org/Archives/Public/www-tag/2002Jul/0302.html

I know this is very anal of me, but when you say:

        "It is the **very same document** that is
         was before I put it on the server, and it even has the **very same
         name** that it had then."

What are the invariants? At some level that file on the server has a physical existence occupying phyical space on a physical spinning platter (or charge wells in some physical other media). As you move it around... you might be making copies/clones - are they all "...the **very same document**..." or are they distinct copies of something - or possibly distinct instance of a class of resource whose representations are 'indistinguisable'. [BTW: I'm quite open to being told that none of this matters].

Taking the hypertext/representation view - we can have multiple hypertext/representation dispensers that yield identical traces in Jonathan's vocabulary (modulo lower level details involving IP addresses and time stamps). One would see each such dispenser as distinct - though equivalent (for most if not all purposes). But... the representation(s) dispensed seems to me to be of something else (not of the dispenser itself). We can choose to view the referring name as referring to the dispenser itself or to the thing that the dispensed representations are representations-of.

My view, FWIW as I explained previously, is to try to view the hypertext/representation dispensers are part of the machinery of the web, its infrastructure - that gives it (the web) the ability to response to questions about the state of URI named resources - to response with wa-representations of the state of a resource, or in the case of resources that it is unable to inspect, to fail or to refer the enquirer elsewhere that *might* yield useful information. "Webbish Objects" or "hypertext/representation dispensers" 'descent' below the surface of the web, and tend not to be the things that we speak-of or name with URI.

> >> Consider: there were files and documents and images all over the planet long before the Web was invented. If
> >> the basic ontology of Web Architectural theory is based on those,  then how on earth are we to explain what changed when the
> >> actual Web came along?
> >
> > ...we gave them 'bigger' names grounded in a uniform namespace? (or
> > at least mapped many of them to the same).
>
> So it was just the URIs that made the Web? Imagine a SciFi scenario in
> which a huge solar flare completely destroys all electronics on the
> planet in one millisecond. Aircraft on autopilot dive into the ocean,
> power grids collapse, international trade and banking ceases,
> telephones don't work, civilization totters... but wait, help is at
> hand! The URIs are still exactly what they were, and they still are
> global *names*, and so all the **naming** is still intact, global
> namespaces and all. So, apparently, the Web still exists!  Great, we
> can all carry on ordering on Amazon and selling on eBay, right? If the
> Web is just a global system of names, then who needs electronics?

:-)) FWIW: my sci-fi scenario is of a mars lander sending back an image a red-rock in which is carved a symbol that looks very like "http://www.ihmc.us/users/phayes/PatHayes."

Ok... So yes there's a whole bunch of engineering and infrastructure that's needed to. We'd be in the same boat if only the relevant specs. survived along with the names. But I thought this was an architectural discussion - the infrastructure and the engineering animate the architecture or may be (conversely) the architecture is post-hoc and explains what the engineering/infrastructure was seeking to accomplish.

AWWW as it was written (circa 2004) has three main sections (roughly): Identifiers; Representations and Formats; and Interaction/Protocol - so there was some growing uniformity across all three of these pieces.

Returning to your question: what happened when the web came along was that some (practical) barriers to publishing information on the internet were drastically lowered; the overall system strived for weak consistency (trading currency for network bandwith with caches) and 'forgiving' behaviours in the face of errors (404s and HTML parsing), some of which are biting us now.


>
> Pat
>
> >
> > Stuart
> > --

Stuart
--
> >
> >> -----Original Message-----
> >> From: Pat Hayes [mailto:phayes@ihmc.us]
> >> Sent: 10 June 2009 15:31
> >> To: Williams, Stuart (HP Labs, Bristol)
> >> Cc: David Booth; Alan Ruttenberg; Jonathan Rees;
> >> noah_mendelsohn@us.ibm.com; AWWSW TF
> >> Subject: Re: Are generic resources intentional?
> >>
> >>
> >> On Jun 10, 2009, at 9:00 AM, Williams, Stuart (HP Labs,
> >> Bristol) wrote:
> >>
> >>> Hello Pat,
> >>>
> >>>> -----Original Message-----
> >>>> From: Pat Hayes [mailto:phayes@ihmc.us]
> >>>> Sent: 10 June 2009 14:29
> >>>> To: Williams, Stuart (HP Labs, Bristol)
> >>>> Cc: David Booth; Alan Ruttenberg; Jonathan Rees;
> >>>> noah_mendelsohn@us.ibm.com; AWWSW TF
> >>>> Subject: Re: Are generic resources intentional?
> >>>>
> >>>>
> >>>> On Jun 10, 2009, at 7:50 AM, Williams, Stuart (HP Labs,
> >>>> Bristol) wrote:
> >>>>
> >>>>> David,
> >>>>>
> >>>>> I think the point of Alan's question is not so much
> about whether
> >>>>> the file (and hence its representations) can be subject
> to change,
> >>>>> but about whether its is the file itself[*] that is the
> responding
> >>>>> entity or the thing (filing system) that acts as its container.
> >>>>
> >>>> Right. But hasnt it been assumed since day one, ie
> somewhere around
> >>>> Roy's thesis, that The Resource is **the thing identified by the
> >>>> URI**, and that the stuff that gets sent (by the
> Resource, when you
> >>>> ping it suitably) is a Representation of it, ie of the Resource,
> >>>> rather than the Resource itself.
> >>>
> >>> Indeed... it was just that David's response seemed to miss what I
> >>> took to be the point of Alan's question.
> >>
> >> I understand, and agree. But was then (in my pit-bull way)
> >> reiterating
> >> what I believe is the main point.
> >>
> >>>
> >>>> So indeed, a bare text file is *not*
> >>>> a Resource in this sense, rather in the way that my cat
> >> cannot answer
> >>>> the telephone, even though you can hear it meowing when I
> >> answer the
> >>>> telephone.
> >>>
> >>> :-)
> >>>
> >>> I think I have previously taken the view that http requests are
> >>> 'questions' one asks of 'the web' about things named by URI
> >> and that
> >>> responses are answers from 'the web'
> >>
> >> OK..
> >>
> >>> and have tried (repeatedly) to avoid having the machinery
> >> of the web
> >>> (servers, proxies, conceptual http endpoints etc) intrude into the
> >>> domain of this discourse - so I'll acknowledge that distinguishing
> >>> between the file and the machinery (file system) that serves up
> >>> representations of it crosses that self imposed line.
> >>
> >> Well, I also don't want to get too mechanical and all involved with
> >> proxies and stuff, agreed. But I don't think that making a
> >> distinction
> >> between what one might call Web-passive entities (files,
> >> images, ...)
> >> and Web-active ones requires us to dive deep into the
> >> machinery. Think
> >> of the distinction between agents and non-agents for the kind of
> >> conceptual level I'm aiming for. And the 'active' thing
> doesn't have
> >> to be a file system or anything that specific. But it has to be a
> >> thing that can, conceptually, **do** something Webbish. If it just
> >> sits there and exists, then there is no way to even make it be
> >> relevant to the Web *at all*. Other, of course, than being
> something
> >> that can be referred to, but then that encompasses everything.
> >>
> >>>
> >>>> Resources have to be able to Do some Webbish things,
> >>>> participate in the Web architectural dance in some way. They are
> >>>> agents, not files.
> >>>
> >>> So... we're back roughly at the nub of Alan's question, roughly
> >>> "...which is it, there resource of which we speak... the
> >>> (passive)document/work of which the server wa-representations are
> >>> representations of; or the (active) agent'y entity that provides
> >>> responses to questions. I believe that the traditional view
> >> is that
> >>> the URI names/identifies/refers-to the document/work thing rather
> >>> than the (conceptual) machinery in the web (which some have dubbed
> >>> http-endpoint).
> >>
> >> Right, sorry I missed that earlier question. Good
> question, but Alan
> >> gives exactly the wrong answer. I'm not sure what tradition he is
> >> referring to, but the REST/tag/awww answer is surely that it
> >> has to be
> >> the active agent'y entity. Consider: there were files and documents
> >> and images all over the planet long before the Web was invented. If
> >> the basic ontology of Web Architectural theory is based on
> >> those, then
> >> how on earth are we to explain what changed when the
> actual Web came
> >> along?
> >>
> >>>> Seems to me that several very smart people worked hard
> to get this
> >>>> broad architecture picture worked out, and that we should use it
> >>>> rather than ignore it.
> >>>
> >>> Certainly... Though I wasn't conscious of ignoring it...
> >>
> >> No, sorry, I wasnt aiming this remark at you particularly.
> >>
> >> Pat
> >>
> >>> though maybe I was. Mostly I was trying to point to what I thought
> >>> was the point of Alan's question which seemed to me to have been
> >>> missed.
> >>>
> >>>>
> >>>> Pat
> >>>
> >>> BR
> >>>
> >>> Stuart
> >>> --
> >>>
> >>>>
> >>>>
> >>>>
> >>>>>
> >>>>> Stuart
> >>>>> --
> >>>>>
> >>>>>
> >>>>>
> >>>>>> -----Original Message-----
> >>>>>> From: David Booth [mailto:david@dbooth.org]
> >>>>>> Sent: 10 June 2009 11:02
> >>>>>> To: Alan Ruttenberg
> >>>>>> Cc: Pat Hayes; Jonathan Rees; noah_mendelsohn@us.ibm.com;
> >>>>>> AWWSW TF; Williams, Stuart (HP Labs, Bristol)
> >>>>>> Subject: Re: Are generic resources intentional?
> >>>>>>
> >>>>>> On Tue, 2009-06-09 at 22:45 +0100, Alan Ruttenberg wrote:
> >>>>>>>> Why not, "can emit a response to some kind of access
> >>>>>>>> protocol"  ? That seems
> >>>>>>>> to handle all the present and all the likely future cases, be
> >>>>>>>> unambiguous,
> >>>>>>>> and (by philosophical standards) vividly clear and
> >> unambiguous.
> >>>>>>>> And it has
> >>>>>>>> the great merit of talking about the **actual
> >> resource** rather
> >>>>>>>> than an
> >>>>>>>> awww:representation of it, which (latter) is what gets
> >> conveyed
> >>>>>>>> in messages,
> >>>>>>>> in fact.
> >>>>>>>
> >>>>>>> What does  "can emit a response to some kind of access
> >>>>>>> protocol"  the answer to?
> >>>>>>> Notably, it doesn't include things like text files with
> >> html in
> >>>>>>> them.
> >>>>>>
> >>>>>> Sure it can.  If you think of these things as functions
> >> from time
> >>>>>> and
> >>>>>> requests to representations then its representations still may
> >>>>>> change
> >>>>>> over time (as the file is modified) even if at any given
> >> time it
> >>>>>> always
> >>>>>> emits the same representation regardless of the request.
> >> Or, if
> >>>>>> you
> >>>>>> take Roy's "curried" view (see
> >>>>>>
> >>>>
> >>
> http://lists.w3.org/Archives/Public/public-awwsw/2008Apr/0047.html )
> >>>>>> of these things being functions from time to
> representation sets,
> >>>>>> then
> >>>>>> even if the representation set is a singleton set at a
> >>>> given time it
> >>>>>> still may be a different singleton set at another time, when
> >>>>>> the file is
> >>>>>> modified.
> >>>>>>
> >>>>>>
> >>>>>> --
> >>>>>> David Booth, Ph.D.
> >>>>>> Cleveland Clinic (contractor)
> >>>>>>
> >>>>>> Opinions expressed herein are those of the author and do not
> >>>>>> necessarily
> >>>>>> reflect those of Cleveland Clinic.
> >>>>>>
> >>>>>>
> >>>>>
> >>>>
> >>>> ------------------------------------------------------------
> >>>> IHMC                                     (850)434 8903 or
> >>>> (650)494 3973
> >>>> 40 South Alcaniz St.           (850)202 4416   office
> >>>> Pensacola                            (850)202 4440   fax
> >>>> FL 32502                              (850)291 0667   mobile
> >>>> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>>
> >>>
> >>
> >> ------------------------------------------------------------
> >> IHMC                                     (850)434 8903 or
> >> (650)494 3973
> >> 40 South Alcaniz St.           (850)202 4416   office
> >> Pensacola                            (850)202 4440   fax
> >> FL 32502                              (850)291 0667   mobile
> >> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
> >>
> >>
> >>
> >>
> >>
> >>
> >
>
> ------------------------------------------------------------
> IHMC                                     (850)434 8903 or
> (650)494 3973
> 40 South Alcaniz St.           (850)202 4416   office
> Pensacola                            (850)202 4440   fax
> FL 32502                              (850)291 0667   mobile
> phayesAT-SIGNihmc.us       http://www.ihmc.us/users/phayes
>
>
>
>
>
>
Received on Thursday, 11 June 2009 11:03:56 GMT

This archive was generated by hypermail 2.2.0+W3C-0.50 : Thursday, 11 June 2009 11:03:57 GMT