See also: IRC log
<DanC_lap> Scribe: Dan Connolly
<DanC_lap> ScribeNick: DanC_lap
<Stuart> - Reviewing "Cool URIs..."
<Stuart> - #fragId / indirect identification
<Stuart> - namespaceDocument-8
<Stuart> - Web 2.0
<Stuart> - Noah's WebArch Presentation
<Stuart> - URI Testing
<Stuart> - WebArch Vol 1 2nd Ed, and/or Vol 2: How to get started?
<Stuart> https://gnowsis.opendfki.de/repos/gnowsis/papers/2006_11_concepturi/html/cooluris_sweo_note.html
HT: we agreed to ask to add "http: " to the "On the Semantic Web, URIs identify ..." , right? [right]
NM: "On the Semantic Web, "
suggests there are 2 webs. I understand this is tutorial, but I
think this concern could be addressed with something like...
[missed]
... "With the addition of Semantic Web capabilities..."
(consensus emerges; Norm is editing a review comments message)
(see also Users/ndw/TAG on norm's disk. @@point to www-tag msg)
NM: "description of a URI" is a use/mention bug... should be "description of a resource identified by a URI"
DanC: not sure it's cost-effective to repair all such use-mention buglets in informal text
SKW: "identified resource" works
DanC: OK
... I wonder if people are using this "Be on the Web" box to
argue against doc#term URIs. Note that "description of" is not
"representation of"
SKW: acme.com is used, still. needs to be example.com
<DanC_> (hmm.. N3 syntax... it actually only uses the turtle fragment; might be better to cite turtle than N3. maybe I'll send that as an individual comment)
<Rhys> DC: Draws diagram on whiteboard. SW takes picture.
<Stuart> > Whiteboard photo archive
<Rhys> DC: Suggests that the story could be told by successive elaboration of the diagram. Starting from the statement about 'Don't be ambiguous' and refining through the various approaches that are proposed.
[scribe missed quite a bit... picking up...]
HT: there's some stuff about
fragments and conneg, no? [indeed... http://www.w3.org/TR/webarch/#frag-coneg
]
... going back to our discussion of namespaceDocument-8 and the
XML Schema datatype namespace document, we're saying that the
difference between a datatype and a paragraph doesn't matter
for this purpose
DanC: umm... yeah.
HT: on the other hand, we get a contradiction if xsd:boolean is a paragraph and a datatype [presuming datatypes and paragraphs are disjoint]. So www.w3.org shouldn't give a 200 for the /2001/XMLSchema namespace URI
DanC: I don't think so; I prefer to ammend the HTML media type so that head/@profile takes away some constraints on fragids
[there are other designs in this space too]
<Stuart> Whiteboard photo is attached to: http://lists.w3.org/Archives/Public/www-archive/2007Sep/0061
SKW: note Dan's diagram uses distinct style for URIs and resources. that seems significant
NM: not clear that "err on the
side of caution" is well-motivated...
... e.g. in the case of a relational table, TimBL and I agreed
that a 200 was ok...
... it's not clear to me that doing a redirection where it's
not needed isn't more harmful than doing a 200 where it
shouldn't have.
SKW: the "all its essential characteristics can be conveyed in a message" is a bit out of context; we didn't give that as definition of Web document, but information resource
DC noodles... ... not well motivated; for example, we think many relational tables are information resources, and this recommendation would result in unnecesary redirections.
<Stuart> http://lists.w3.org/Archives/Public/www-tag/2007Sep/0017.html seems relevant
SKW: so NDW has an action to
finish this review...
... meanwhile, what about the hypothesis that "Cool URIs..."
obviates much of Rhys's draft on HTTP URIs?
DanC: in that my main goal is to help people choose URIs, yes.
NM: with some reservations about other HTTP status codes and whether we're elaborating or suggesting changes to the HTTP spec, yes
Rhys: I think "Cool URIs..." addresses much of what "Dereferencing HTTP URIs" was going to say, but there are some other bits that seem useful for covering other aspects of ISSUE-57/ HttpRedirections-57 and ISSUE-28 (fragments)
HT: to some extent, yes [scribe
struggles to capture the gist of what he actually said]...
though there are some critical bits around "web presence" of
Rhys's draft that I want us to work on
... I'm somewhat sympathetic to the point that a write-up of
our decision on httpRange-14 is still in order.
DanC: looking at the redirections ISSUE-57, I see bits like "Particular concerns are the cachability (or otherwise) of HTTP redirection responses" that I don't think "Cool URIs..." should cover that.
Rhys: quite
... one possibility re more-than-a-mail-message on issue 14,
perhaps webarch 2nd ed?
HT: well, but I'm not holding my breath
<scribe> ACTION: Norman relay comments as constructed today to the SemWeb EO IG editors of "Cool URIs..." [recorded in http://www.w3.org/2001/tag/2007/09/19-tagmem-minutes.html#action01]
<trackbot-ng> Created ACTION-52 - Relay comments as constructed today to the SemWeb EO IG editors of \"Cool URIs...\" [on Norman Walsh - due 2007-09-26].
trackbot-ng, ACTION-52 is on ISSUE-57
<Rhys> PROPOSED RESOLUTION: TAG will not publish the Dereferencing URIs draft in its current form. We anticipate that a suitably updated version of the Cool URIs document will provide appropriate guidance on the httpRange-14 finding.
DanC: we don't need decisions to not do something, but updating the status seems worthwhile
trackbot-ng, status
e.g. http://www.w3.org/2001/tag/group/track/users
no, that page has rlewis2
<Rhys> ACTION: Rhys to update the current version of the Dereferencing URIs draft to show that we will not formally publish it. [recorded in http://www.w3.org/2001/tag/2007/09/19-tagmem-minutes.html#action02]
<trackbot-ng> Created ACTION-53 - Update the current version of the Dereferencing URIs draft to show that we will not formally publish it. [on Rhys Lewis - due 2007-09-26].
BREAK for lunch, 'till 1:15pm
<Rhys> Scribe: Rhys
<scribe> Scribenick: Rhys
<scribe> ACTION: Norm to produce a new draft of the namespace documents finding based on 18 Sep discussion (ACTION-56) [DONE] [recorded in http://www.w3.org/2001/tag/2007/09/19-tagmem-minutes.html#action03]
<ht> I would change this sentence "The nature key is the label which allows us to distinguish between the different targets that could be used for the purpose. @
<ht> to
<ht> "The nature key is the label which allows us to distinguish different contributions that the same resource can make"
NW Projects draft document
DC: Asks about the full URIs for
various thinkgs, such as the nature key
... The proposal is that for RDDL, they will own the URIs?
NW: Yes
SW: Use of term key occurs earlier than it's definition.
NW: that's an editorial issue to
be fixed.
... Draws diagram on whiteboard showing the relationships
between namespaces, natures, ancilliary resources.
DC: Trying to determine which of this is the generic model and which is specific to RDDL
RL: Notes that the discussion is about Henry's example, based on RDDL
DC: Maybe the title of section 2 should be 'using RDDL'?
HT: But it's not about RDDL, we don't mind where the RDF comes from.
DC: Do we need to call this 'the model'?
HT: I think that we need to do this to address the concerns previously expressed to the TAG
NW: A couple of years ago, we decided that we couldn't change RDDL, but that we could create a model for an abstraction that would let us represent this information.
DC: The problem is that calling it 'The Model' is very definitive
NM: Maybe we should look at the parts of this that are mandatory.
HT: This part of this document is not about what you must do, it's about the resources that are available to you when writing your namespace document.
NM: We do say that if you create
a namespace we are recommending that you have a namespace and
that you have some materials on the web.
... We could say, that you need to put up some stuff about the
namespace, and that, whatever the mechanism (RDDL or whatever),
there is a set of information that you should have for the
namespace
HT: I don't agree about standardising what should be in the namespace document. RDDL provides a vocabulary for imparting the information to humans. We are extending this to machines
NM: I thought we were defining
the way that you could get something that could then be used to
get back to RDF. I thought we were trying to define a preferred
way to get you to a schema.
... Could be stylesheets, schemas etc.
DC: What about using RDF datatypes?
HT: This is about connecting resources, not datatypes
SW: This part of the narrative is about ancilliary resources to be used with the namespace. There is another set of information about the things that are in the namespace.
HT: As of today, if you go to the
namespace document, you'll find just connections to associated
resources.
... However, there are assertions in the old namespace document
about links to names in namespaces but those anchors don't
exist
NM: So we are moving in the direction that the new description document has both connections to associated resources and datatypes
HT: So the example includes definitions that use RDF datatypes and relate to the names in the namespace
NM: Trying to infer what we can
do in the general case. Is it correct to have the individual
names in the same document as the associated resources. It's
probably OK.
... Maybe should be a standard way in all RDDLs of referring to
a name in the namespace.
<DanC_lap> (RDDL's normative reference competes with rdfs:isDefinedby. )
HT: I think that you could
conclude that the schema #boolean is either a name in the
namespace or an associated resource or both.
... There are two ways to improve this. Either in every
datatype statement, could add an additional RDF type named
thing (thing identified by something in the namespace).
DC: Usually this is done by rdf is-defined-by
NM: Whatever the appropriate expression of this, it seems as though a triple would be the right relationship
DC: In practice, the processors look for a schema and then carry on processing.
HT: Just because x#y resolves, doesn't mean that y is a name in the namespace.
NM: my sense is that I'm asserting that the name I've defined is in the namespace
DC: In practice the way that people do this is what Henry wrote on the board.
<DanC_lap> which is: { xsd:boolean rdfs:isDefinedBy "" }
NM: To me the namespace is a set of names.
NW: There are a couple of puns going on here, so it works out.
NM: I see the namespace is an information resource. The associated document is a description of the namespace.
HT: If the W3C server were a standard Apache server, and you requested the XML Schema URI, you'd get a 404. I'd be happy if when you retrieved schema.html you got a 200, and when you asked for the schema you'd get a 303 to the HTML document and if you asked for namespace as RDF, it GRDDLs the namespace html document and returns the RDF.
NW: You'd have to use xml:base in
the RDF.
... Henry's description is a little different to the 303 story
we have been telling so far.
NM: Often the representation is a
rather 'noisy' representation of the resource (advertising
etc.)
... Since we are used to that, it seems less of a stretch to
claim that Henry's story covers representations of the
namespace
... I don't think that we don't need the 303
DC: Possible argument for the 303. Dan draws diagram. Shows URI of a namespace. Access to the namespace URI 302s to the HTML document. If you ask for RDF you get a 200 and an RDF representation, possibly from the GRDDL.
<Stuart> 10.3.3 302 Found
<Stuart> The requested resource resides temporarily under a different URI.
NM: Why not return HTML by conneg from the namespace URI
HT: To preserve the difference between the namespace and namespace document
SW: I think 302 is wrong, because
it is not temporarily moved.
... I think it should be 301.
NM: I don't agree, because 301
means it's permanently moved.
... I think it's 303. But I also feel there is an asymmetry.
Both the RDF and the HTML have the same relationship to the
namespace URI.
HT: If today, with the magic in the configuration on the W3C site, you request HTML from the namespace URI, you get it, but you also get a content header.
NM: What is the implication of content location header?
DC: You get the representation, but it's also a representation of this other URI
<Stuart> 14.14 Content-Location
<Stuart> The Content-Location entity-header field MAY be used to supply the
<Stuart> resource location for the entity enclosed in the message when that
<Stuart> entity is accessible from a location separate from the requested
<Stuart> resource's URI. A server SHOULD provide a Content-Location for the
<Stuart> variant corresponding to the response entity; especially in the case
<Stuart> where a resource has multiple entities associated with it, and those
<Stuart> entities actually have separate locations by which they might be
<Stuart> individually accessed, the server SHOULD provide a Content-Location
<Stuart> for the particular variant which is returned.
NW: I'll make the changes to the
diagrams, and then hand the words over to you Henry.
... Claims that his action on this document is now complete
<DanC_lap> trackbot-ng, status
<scribe> ACTION: ht due in 2 weeks produce another revision of the namespaceDocument-8 draft from Norm [recorded in http://www.w3.org/2001/tag/2007/09/19-tagmem-minutes.html#action04]
<trackbot-ng> Created ACTION-54 - Do in 2 weeks produce another revision of the namespaceDocument-8 draft from Norm [on Henry S. Thompson - due 2007-09-26].
<DanC_lap> http://esw.w3.org/topic/UriTesting
DC: URI mailing list doesn't
correspond to any working group at W3C. TAG monitors the
list.
... I started a Wiki page on URI testing. People copy and paste
each others testing techniques. There is currently no formal
coordination.
... Tempted to suggest that some kind of interest group be
established on URI and IRI testing.
... There is a new Java API in JCP for performing URI
manipulations separate from the actual HTTP operations
... I think the time is right to organise something around
this.
... I'd like it to be relatively disconnected for the same
reason as to keep the manipulations separate from HTTP
SW: Dan, what would you like the TAG to do?
DC: We could say that this is a good idea and ask the director to consider starting some kind of activity.
General discussion about what the form of any such follow on activity might be
<DanC_lap> ACTION: Dan work with SKW on a few paragraphs of thinking around a URI testing group (IG/WG/XG?) [recorded in http://www.w3.org/2001/tag/2007/09/19-tagmem-minutes.html#action05]
<trackbot-ng> Created ACTION-55 - Work with SKW on a few paragraphs of thinking around a URI testing group (IG/WG/XG?) [on Dan Connolly - due 2007-09-26].
<DanC_lap> http://www.w3.org/2001/tag/group/track/actions/2
DC: Where are we on the action?
<Stuart> http://lists.w3.org/Archives/Member/tag/2007Sep/0012
<ht> http://www.w3.org/2001/tag/2007/04/30-minutes is relevant
NM: I've not had a response. I'll try again.
<DanC_lap> ACTION: NM contact Brutzman etc... ACTION-2 [CONTINUES] [recorded in http://www.w3.org/2001/tag/2007/09/19-tagmem-minutes.html#action06]
<Noah> Slashdot link to standardization effort for virtual worlds protocols: http://games.slashdot.org/games/07/09/19/126250.shtml
<Noah> Contains link to 2nd Life Wiki: http://wiki.secondlife.com/wiki/Architecture_Working_Group
DC: We've not followed up with Croquet
NM: I think that Croquet is more
of a distributed environment for professional use
... Do you build your own community?
DC: Tend to have your own. They are looking at more global communities
HT: QWAQ is trying to build a business model on top of Croquet.
SW: About to close. 3 days was good. Enjoyed the group work and the progress has been good.
NW: 3 days is good.
General agreement
NW: Don't assume that just because we don't have a third day's agenda we only need a two day meeting
NM: Liked the ability to get some small stuff done on the third day.
General feeling that open agendas can be useful, but we shouldn't necessarily assume this is always a good idea.
SW: We are adjourned
HT: I've just found that in Croquet, the connections between rooms are called portals. These are labelled with TPostcards that are not URIs. However, they are described as being similar to URLs in web browsers.