GRDDL Weekly -- 14 Feb 2007

<DanC> heeere Zakim

<HarryH> Anyone *want* to be scribe?

<HarryH> HarryH: asking for someone to scribe

Convene GRDDL WG meeting of 2007-02-14T11:00-0500

<HarryH> scribe: JohnClark

HH: Regrets from Fabian
... and from Chime

<HarryH> PROPOSED: to approve GRDDL WG Weekly -- 07 Feb 2007 as a true record

<HarryH> APPROVED: http://lists.w3.org/Archives/Public/public-grddl-wg/2007Feb/att-0088/07-grddl-wg-minutes.html minutes 7 Feb

<HarryH> Zakim +Ian_Davis is the true record

[#issue-http-header-links]

<danja> thanks Harry - you're a quicker typist ;-)

<HarryH> * ACTION: IETF Link and Profile Headers to be included as "feature at risk", and dropped if they are not approved by IETF by time of move to Proposed Recommendation.

HH: This is my action
... keep this as a feature at risk with a strict timeline?

<HarryH> http://www.w3.org/2001/sw/grddl-wg/

<HarryH> ACTION: HarryH to e-mail Ivan to double-check this. [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action01]

<HarryH> The question is whether to accept a schedule risk to add a 2 month candidate recommendation phase.

<HarryH> before going to proposed recommendation.

<HarryH> PROPOSAL: Is to keep the headers and add a 2 month candidate recommendation phase.

<HarryH> PROPOSAL: Is to keep the headers but do not add a 2 month candidate recommendation.

<HarryH> Any opinions?

<DanC> (I don't think the header can be done without the 2 month CR phase.)

Murray: If the headers are sound, then we should include them in the spec

<HarryH> bwm: No dependencies in the way.

Murray: What happens if we don't include this feature?

HH: We go straight from CR to R

Murray: Another risk is backtracking based on feedback if we stick with the feature

Ian: Likes the feature, as discussed
... If the IETF process doesn't complete, it's a small amount of work to take it out

GRDDL Spec: Last Call

HH: Do the current tests cover RDF/XML docs well enough?
... Other comments addressed?

??: Mine were minor

<DanC> ACTION: DanC to address rreck's comments in http://lists.w3.org/Archives/Public/public-grddl-wg/2007Feb/0042.html [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action02]

John: Thumbs up on the spec

danja: Thumbs up ...

DanC: There is still stuff not done.

HH: Can it be done soon?

DanC: Yes, but they're normative.
... Particularly see Section 5.
... Does the GRDDL profile get processed by this mechanism?

HH: What is the difference between this and the namespace mechanism?

DanC: You don't follow your nose to the GRDDL profile URI in this case

--- questions arise on how namespace documents work

DanC: PO document, GRDDL agent goes to NS doc which says "world is flat"; is that part of a GRDDL result for the original doc?

John: Isn't that the way the spec works?

DanC: Nope

<DanC> currently the GRDDL spec *only* lifts namespaceTransformation triples from namespace documents.

But doesn't it then use those to create GRDDL results from the original doc (that could include "world is flat")?

DanC: We've agreed that section 3 is ok.
... but there is still a question about section 5
... What happens when there are multiple profiles?
... Does this question even matter?

<DanC> jjc's comment http://lists.w3.org/Archives/Public/public-grddl-comments/2007JanMar/0040.html

<DanC> "because I don't have a hard coded

<DanC> rule for profileTransformation, but rely on the GRDDL transform of the

<DanC> GRDDL profile."

DanC: There may be some implementor's confusion here
... I need to go over it again to make sure this mechanism is sound.

Crap.

<bwm> I just sent mail that I'm satisfied that issue-mt-ns has been addressed.

<DanC> looking for eRDF details... http://esw.w3.org/topic/CustomRdfDialects ...

<bwm> http://purl.org/NET/erdf/profile

<DanC> http://purl.org/NET/erdf/profile

<DanC> <p>This following link provides the statement : <a

<DanC> rel="profileTransformation"

<DanC> href="http://purl.org/NET/erdf/extract-rdf.xsl">extract-rdf.xsl</a>, <a

<DanC> rel="profile" href="http://purl.org/NET/erdf/profile">profile</a></p

<DanC> <link rel="transformation"

<DanC> href="http://www.w3.org/2003/g/glean-profile" /

DanC: May want a standard library of transformations

<DanC> ACTION: Ian to clarify profileTransformation for JJC [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action03]

<HarryH> "if an information resource ?D has an XHTML representation whose profile attribute refers to ?PROFILE, then any GRDDL result of ?PROFILE is a GRDDL result of ?D"

HH: Remove that from the spec?
... Do we have enough guidance to sort this issue out?

<DanC> possible new issues: (a) whether rules in section 5. GRDDL for HTML Profiles apply to http://www.w3.org/2003/g/data-view

<DanC> (b) whether http://www.w3.org/2003/g/glean-profile should be in the standard library

<DanC> possible tests for "no, namespace documen triples aren't part of grddl results"

<DanC> and likewise profiles

DanC: We're comfortable with the namespace document mechanism?

<DanC> possible test for exactly which triples are in grddl results of http://www.w3.org/2003/g/glean-profile

<HarryH> So, separating the transformation triples from the namespace, not a bug but a feature.

<HarryH> ACTION: danja is sketches the test in e-mail and bwm checks in it, but not required in last call. [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action04]

<DanC> which test being "no, namespace documen triples aren't part of grddl results"

HH: We don't have consensus on last call due to JJC's security issues and namespace document questions

DanC: Do we want to reopen discussion on the "GRDDL-aware agent" conformance label?

<danja> "GRDDL-aware agent" = "a

<danja> software system which implements the mechanisms described in this

<danja> document".

DanC: "The GRDDL marketplace want to shop for 'GRDDL-aware agents'"
... but I'm concerned about the costs, which have largely been dealt with by JJC's proposal

IanD: Can we incorporate JJC's proposal as implementation advice instead of using a CL?

<DanC> indeed, "GRDDL-aware agents MUST support XSLT1" is something that might come up

HH: will we have to tighten the spec if we use a conformance label?

<briansuda> abstain

DanC: poll: conformance label, or no?

<rreck> i vote for conformance labels

<DanC> y, a a, n, y, ...

Murray: spec is good right now, but a CL would be a valuable tool in the marketplace

<DanC> y, a a, n, y, y/n,

<DanC> y, a a, n, y, y/n, y/n...

DanC: Status quo is "gee, this could be dangerous"

HH: JJC's proposal gives specifics, and would require a conformance label to enforce

<HarryH> \me and the question is should we write "You MUST do this to call yourself a GRDDL"

<danja> aside: in practice a GRDDL-aware agent may return different results for a given source document at different points in time (e.g. downstream profiles/transforms may 404)

Murray: isn't there a risk that such a change would require more time?

DanC: I think there's a good chance this could be done quickly

HH: And we return to the idea of not including conformance labels

Murray: - wants to see what Dan comes up with
... What about a separate document defining a "GRDDL-aware agent"?

HH: ... make a tech report out of the test suite, with conformance labels there
... risk of losing momentum

DanC: But implementations are proceeding regardless

<HarryH> ACTION: DanC write the draft "conformance label" text, and then johnL and rreck and bwm (jeremy) will then re-read spec. [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action05]

HH: Other open issues?

<Zakim> DanC, you wanted to note section 5 GRDDL for HTML Profiles isn't done

Test cases for GRDDL (Perhaps needed for Last Call)

HH: created RDF/XML test cases

DanC: marker for "maximal" GRDDL result?

HH: JJC provided HTTP header test case
... What if links in headers and in doc conflict?

<DanC> ACTION: DanC to add GRDDL header spec section [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action06]

DanC: Spec still needs HTTP header text

<HarryH> http://lists.w3.org/Archives/Public/public-grddl-wg/2007Feb/0057.html

HH: What has priority? Doc, or header...

DanC: Don't want "priorities"

bwm: Aren't they additive?
... working on getting the test into the test suite

<HarryH> ACTION: bwm (jeremy) to put http header test in test suite [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action07]

<DanC> bwm, pls make a subdirectory under /td/ before you add the .htaccess, ok?

Primer Document: Going To Last Call?

HH: Use N3 or not?

danja: Rec track for this?

HH: Straw poll said yes

<HarryH> Rec track for Primer, it's easier to find.

HH: (easier to find)

<DanC> (er... what's with FROM >http://www.w3.org/2001/sw/grddl-wg/doc29/david-erdf.rdf< ? that's bogus SPARQL syntax)

<rreck> i concur. RDF/XML

danny: Found primer hard to read

<DanC> (have I got some sort of browser bug? "A diagram indicating the sequence of steps described for obtaining RDF from a document using an explicit link to the transformation as described in the preceding paragraph")

danja: Only blocking problem is the XFN RDF/XML currently in the text

rreck: Submitted changes, haven't heard back yet

HH: Willing to fix easy problems

<HarryH> ACTION: HarryH to check those primer fixes in. [recorded in http://www.w3.org/2007/02/14-grddl-wg-minutes.html#action08]

Use-Case Document: Going To Last Call?

<briansuda> nope, i sent my suggestions

<HarryH> briansuda, happy with use-case document?

danja: Thumbs up, generally; suggestions submitted

<briansuda> i'm happy

<HarryH> OK.\

<HarryH> Meeting Adjourned.

Is that enough to turn Zakim off?

<rreck> bubye

<danja> bye

- DRAFT -

GRDDL Weekly

14 Feb 2007

Attendees

Contents