See also: IRC log
Stuart: Are we ok with Jacek observing?
<Stuart> jacek, please intro your self for the record
<observer_JacekK> I am here as the chair of SAWSDL WG and member of WS-Desc WG
<observer_JacekK> Jacek Kopecky, DERI Innsbruck
Stuart: whenToUseGet-7 has had
some discussion, pointed to from the agenda.
... For xmlFunctions-34 I'm hoping Henry will join
... Someone brought up CURIEs on the member list.
... Rhys asked about AOB.
Rhys: Yes, a finding for httpRange-14.
Stuart: Everyone happy with that agenda?
Raman: Passwords in the clear seems to have slipped through the cracks.
Stuart: Yes, I'll continue to coordinate with Ed until he can come.
Henry to scribe.
Regrets from TimBL.
Stuart: I put out an appeal for
topics to discuss; several have responded.
... I'd like to do a little bit of reflection at the beginning of the meeting.
... Noah has suggested that we spend some time on the self describing web.
Noah: You intend that to include both the broader issue and xmlFunctions?
... Norm will lead something on namespaceDocument-8
... Dan/Raman both indicated interest in some discussion on tagSoupIntegration-54
... I hope Norm and Dan/Raman will fill in that part of the agenda.
<DanC> (the version of the acme thing from last ftf is world-readable somehwere)
Stuart: Can we make progress on semantic web architecture at the f2f?
TimBL: I don't think we can make progress on it for this f2f.
Norm: Alas, nor will I.
Stuart: Ok, I'll take it off the agenda for this f2f.
Noah: I'm happy with some open time on the agenda so that we can have deeper discussions.
Stuart: That would leave substantive topics: self describing web, namespaceDocument-8, and tagSoupIntegration-54.
Raman: I'd like to make sure that some deliverables come out of the meeting.
<Noah> Why are our deliverables not Findings, etc.?
DanC: Are you offering?
Raman: If we can define what it
is, I might offer.
... A set of clear notes that we publish is something valuable.
... Notes that only appear in the minutes is not as useful.
<Noah> We could consider someday publishing a AWWW Version 2, etc., but I see no need for a concrete deliverable from each F2F. For me, the F2F is means toward the other ends.
Raman: I'll write something if you'll let me interrupt periodically to get conclusions.
TimBL: There's a big overlap between the self describing web and the semantic web. Noah: do you plan to cover GRDDL and other technologies?
Noah: The short answer is yes,
but the question is how deep to go.
... On the XML side, we have Henry doing xmlFunctions.
... The self describing web feels like an important topic, but I'm not sure I have it crisply defined in my mind.
... So I'm hoping that we can come to some kind of answer to these questions at the meeting. But yes, I was sort of thinking we'd cover all of these things.
... What's nice about RDF and XML is that you can inspect them in deeper ways than you can, for example, a JPEG image.
TimBL: So we might have several
findings: self-describing XML, self-describing RDF, etc.
... I think it would be nice if we could take the self-describing semantic web stuff and explain how the pieces fit together something like we did for the WebArch document.
... Dan, do you think it would make sense to go for a finding that's specifically about self-describing RDF documents, using GRDDL for example? I don't know if we need something to go around the GRDDL spec?
DanC: On the one hand, there's a
sort-of draft finding, a GRDDL rationale document.
... But I missed the hop from self describing web to GRDDL
TimBL: An important part of the SemWeb architecture is that given a URI you can get some triples out of it.
<DanC> (GRDDL background/rationale, perhaps could be a TAG finding when it grows up: http://www.w3.org/2004/01/rdxh/specbg )
TimBL: What makes it
self-describing is that all you need is the URI. Explaining
this requires the URI plus some specifications like HTTP and
... GRDDL seems like the self describing web to me, it's sort of the bootstrap.
Noah: Yes, but I am trying to keep the exposition of the general principles separate from the exposition about specific technologies.
<timbl> "self describing web" ~ "you get stuff from just a URI" which in sem web is "you get triples from just a URI" which involves GRDDL
Noah: One of the important kinds of self description is: can you get triples from this document? Two ways: is it already in RDF or N3; but if the document is in another form, then it's a good thing if there's a bridge from that form to more triples.
<DanC> (what noah said makes sense to me; it seems a small part of the self-describing web story, but if that's what tim had in mind, very well)
TimBL: From the agenda point of view, I was wondering whether the self describing web agendum included GRDDL. Maybe we can continue this discussion in that slot.
<timbl> "self describing web" ~ "you can folow your nose to find out what things mean" -- Ontologies as backing to RDF data including themselves
Stuart: Any other questions about the face-to-face?
<Zakim> ht_london, you wanted to ask about peer-to-peer
Henry: I've arranged to go to MS
Cambridge to give a general sort of URI/webarch talk,
... Mostly I'll be talking to folks interested in the future of peer-to-peer.
... What are we doing about that?
<DanC> (neither "p2p" nor "peer" occurs in http://www.w3.org/2001/tag/issues )
Henry: This is information gathering for me. If there are questions that people would like me to ask folks who are thinking about p2p all day, every day, please send me mail.
Noah: I think we agreed to put schemeProtocols-49 on the back burner.
Henry: I'm not pushing, I just know that it's something we had planned to talk about.
Stuart: These can certainly be topics for breaks and lunch.
<Noah> I still think the schemesProtocols-49 question is an interesting one, but got pessimistic that I could in bounded time tell a story that would garner consensus.
Rhys: I'd be interested in a little discussion of versioning, more from the point of view of where we are and what we need to do
Stuart: Yes. I haven't put it on
the agenda because we have regrets from David who's been
leading that item.
... Many moons ago, the TAG raised a concern about how agents and bits of software that use POST where supposed to know that the operations were safe.
... We had discussions with WSDL WG. They added an attribute to indicate that an operation was safe.
... More recently, there's been work on semantic annotations of WSDL
... The observation has been made in Jacek's question that it would be quite feasible to use the semantic annotations to mark operations safe.
... The message that Jacek sent us offers four proposals and asks for our opinion.
... Jonathan Marsh responded, expressing his preferences.
DanC: The wsdlx:safe is fine by me.
DanC: The <p> tag is a semantic annotation too, but I don't think it's best to indirect via new mechanisms for it.
Stuart: I guess we're being asked if we think there's an architectural reason to do one instead of the other and if we mind if they make that change.
DanC: The WSDL group wants to make this change?
Stuart: No the WSDL group does not want the change.
Noah: Are there two or three options here?
DanC: I see two: keep it or drop it and replace it with a semantic annotations.
<Stuart> from Jonathan's message: 3a. WSDL drops the definition of safety and the dependency upon it
<Stuart> from the HTTP binding, SAWSDL adds it in the normative spec.
<Stuart> The WSDL primer continues to recommend when to use GET or POST,
<Stuart> reinforce the importance of safety, and provide examples and
<Stuart> pointers to the SAWSDL spec.
<DanC> that's 1. the other is: keep it as is.
Stuart: I think my own preference is that the functionality does not disappear, but I don't care how.
DanC: I thought wsdlx:safe was to some extent deployed. If it isn't then I guess it's a coin toss.
Noah: I don't have strong
thoughts on this. Jonathan raised some points that I only sort
of partially understood.
... As long as the annotations meet the need that the WS community thinks it has, and as long as they're clearly defined, I don't feel strongly.
Stuart: So is our position: we want the functionality to be preserved, but we don't think there's an architectural argument.
DanC: That's not my position. I think the mechanism should be as convenient as possible.
Stuart: Of the two, which do you think is more convenient?
DanC: The wsdlx:safe attribute seems more convenient, but it's only more convenient if its actually deployed.
<Noah> I note that the primary utility of the wsdlx:safe annotation is not to
<Noah> default the binding to GET in the absence of other information (there are
<Noah> two other ways to explicitly set the operation to GET which take precedence
<Noah> over wsdlx:safe). The primary utility is to mark a POST operation as safe,
<Noah> as POST may be a preferable method for reasons other than indicating safety.
<Noah> There is no dependency between the HTTP binding and wsdlx:safe in this case.
<Noah> Above is from Jonathan's note.
Noah: Apparently wsdlx:safe is not more than a very gentle suggestion to use GET. The primary utility is simply to mark a POST as safe.
DanC: Does anyone have direct experience with GUI tools for making WSDL interfaces?
<Noah> So, I'm a bit confused about why this all matters if it's mostly being used with POST in practice anyway.
DanC: What I want is a button
that says "make it safe" and if it has less than, say a dozen
parameters, then it gets mapped to GET.
... So a stock quote based on one four character parameter would get mapped to GET.
<Noah> Yes, I suppose I could imagine tools that got some value out of the fact that the post was idempotent, but it seems at best a very small step toward getting the Web to be used as intended.
Stuart: So I don't think we have an answer to their question as such.
<observer_JacekK> it matters because there will be more annotations and inventing WSDL extensions for them is not a good way forward, that's why we have SAWSDL
Stuart: Jacek is concerned about having an approach that is more architecturally clean. I think there's tension with people in the WG with respect to deployment.
DanC: I think that common things should be easy with a small amount of markup.
Stuart: I think I can summarize what we discussed, but I don't think we're making a decision for them.
<observer_JacekK> the button GUI functionality can work with either
TimBL: Yes. I don't think this is
a case where there's a strong architectural principle.
... If it gets moved, who actually has to put it in the spec?
<DanC> " there will be more annotations and inventing WSDL extensions" really? who's building them observer_JacekK ?
Stuart: I think the two WGs are in lock step so it won't get dropped.
<observer_JacekK> I think this is about using the right specs where available, and SAWSDL is available now to WSDL
DanC: I think there's a huge risk. I don't think the folks building GUI tools are connected to the SAWSDL WG.
<observer_JacekK> DanC, we (DERI) build some annotations
<observer_JacekK> and it's the assumption behind the existence of SAWSDL that there will be annotations, but I know this is weak
TimBL: Obviously one of the impacts of putting it in the SAWSDL document is that some tools won't implement it. Is it true that anyone interested in the safe status will be interested in other annotations? It doesn't seem that way to me.
Norm: That sounds like an argument against the annotations mechanism.
<observer_JacekK> argument against SAWSDL or against safety with sawsdl?
TimBL: Yes. The argument for it is, if you've got an annotation mechanism and this is an annotation, you should use the annotation mechanism. There's not an overriding principle. It's a question of what's going to allow it to be implemented.
<Stuart> the latter
Stuart: I think the action is with me to respond to Jacek and Jonathan.
<scribe> ACTION: Stuart to respond to Jacek and Jonathan wrt whenToUseGet-7 and WSDL. [recorded in http://www.w3.org/2007/02/26-tagmem-minutes.html#action01]
<observer_JacekK> I don't know of anyone who uses safety annotation now; I plan to do so in connection with more annotations
<timbl> These are trade-offs. The WG is in the best position to make an educated assesment of the best way to go.
Henry: Last time we met, we talked about xmlFunctions at some length. The document that I produced attempts to reconstruct the core of that discussion.
Henry has problems with the phone; item postponed until the face-to-face.
Stuart: This hasn't surfaced on www-tag.
Stuart: We haven't been tracking it with it's own issue. It seemed worth asking if we want to track it or not.
Norm: I brought it up because it
came up at the XML-CG.
... We don't seem to have reached any conclusions.
Stuart: It's not clear whether the CURIE draft is a public document.
DanC: It doesn't seem to be.
TimBL: They're changing the
definition of the href field in HTML's <a>.
... One issue is you can't mess with that without asking for everyone's permission.
... There are also issues with the CURIE design. The namespace prefix and the URI prefix are overloaded.
... Square brackets are used to disambiguate this when it could be an issue, but there's no way to know when it could be an issue.
TimBL: There are at least those two issues.
Stuart: We've had some dialog on
this at the AC meeting.
... This isn't actually new. But we haven't really been trying to steer a particular direction with it.
<ht_london> There are a lot of changes since the Edinburgh discussion
Norm: I think that we should comment on the draft when it becomes public.
DanC: I sent a comment on my own behalf.
<Zakim> DanC, you wanted to note use of curies in the RDFa draft http://www.w3.org/TR/2006/WD-xhtml-rdfa-primer-20060516/#id69088
Stuart: Should I ask the chair what the status of the drafts are and encourage them to make them public?
Henry: I misunderstood a pointer
to an editor's draft to be a pointer to a public draft.
... Now that we're keeping our eyes on this, I'll wait until there's a public draft.
... I think Stuart should do what he said, noting that the TAG doesn't feel comfortable commenting on a private draft.
<scribe> ACTION: Stuart to contact the Semantic Web Deployment and HTML WG chairs for an update on the status of this document and to encourage them to make it public. [recorded in http://www.w3.org/2007/02/26-tagmem-minutes.html#action02]
Rhys: This could easily be a
newbie question. The traffic on httpRange-14 last week made me
go looking for a finding and I couldn't find it.
... If not, should we create one?
... If we should, I was going to volunteer to write it.
DanC: We haven't made a finding.
There was a resolution and a group decision and Roy announced
... That caused the discussion to decrease and we've treated it like a political third rail since.
Norm: I hadn't earlier argued for a finding because I hadn't tried it. I've now tried it and I'd be more inclined to have a finding, but I'm not willing to write it :-)
TimBL: Other groups have written
... This is also about the self-describing semantic web.
... It's a key part about a how to get from a URI to triples.
Noah: I've always wanted to absorb httpRange-14
TimBL: I think it would be nice
to have TAG document that said, given a URI, this is what Dirk
does with it.
... Some URIs get 303, some 302, etc.
... Fills in the gaps between all the specs.
Noah: Yep, that makes sense.
Stuart: So that's bigger than a finding.
TimBL: Well, what we did last time was take findings of various sizes and try to collect them into an architecture document.
Noah: Yes, but if the self
describing web goes in the direction we think, you're saying
there are things we can point to.
... That fits the larger picture about how you follow your nose.
Rhys: Clearly these things are related, but I still wonder if we need a separate finding on httpRange-14.
<Noah> I think we already have said something about httpRange 14, I.e in Roy's email describing our decision.
<Stuart> see: http://www.w3.org/TR/swbp-vocab-pub/ from SWBPD-WG
TimBL: I wonder if we should work on self describing semantic web documents instead.
Norm: I don't see why it's wrong to have a finding that addresses that single issue and we should let Rhys do it.
<Noah> If we think it's important to either give more detail or wider publicity than Roy's note, fine, but only if the likely value is high. There is a real risk that once we start, lots of time will be burned going over old disagreements. Or not. We might get lucky.
TimBL: Well, one reason to give it a different title is to avoid the third rail effect. But also we've come farther now, so it's more valuable to tell the broader story.
Stuart: The semantic web best
practices not is quite good.
... I'm hearing some note of caution about writing on httpRange-14 and some enthusiasm.
TimBL: I think it would be good to have a document we can point to when people ask about it but give it a different title.
Stuart: Rhys, are you happy to noodle on that and see what you come up with?
Stuart: So we'll let you work on that in your own time for a while.
Rhys: I'm happy to take an action.
<scribe> ACTION: Rhys to consider and draft a finding around the issues raised by httpRange-14. [recorded in http://www.w3.org/2007/02/26-tagmem-minutes.html#action03]
Noah: I'm just about finished with the slides for the enterprise workshop.