See also: IRC log
<chimezie> whats that noise?
<HarryH> scribe: BenA
<HarryH> PROPOSED: to approve GRDDL WG Weekly -- 20 Dec 2006 as a true record
<scribe> ScribeNick: benadida
consensus on minutes... approved
<DanC> Zakim ??P32 is BrianSuda
scribe next week: DanC, who will show RonR for a future date
<HarryH> Withdraw XInclude Test?
HarryH: we might have to withdraw
the XInclude Test
... the test suite could have two outputs, or withdraw xinclude test
... one that does xinclude, one that doesn't
DanC: that's substantive discussion of issue.
HarryH: I thought we had consensus.
<HarryH> RESOLVED: to add issue faithful-infoset (split it out of issue-mt-ns) and close it a la: The GRDDL WG does not mandate XInclude and further XML processing on the source document before the transformation to RDF, and write a caveat in the specification recommending XPROC when this is the case. (benA and BrianS abstaining)
<HarryH> Is this not resolved?
DanC: my understanding is that Xinclude should not happen in the test suite
<DanC> "RESOLVED: to add issue faithful-infoset (split it out of issue-mt-ns) and close it a la: The GRDDL WG does not mandate XInclude and further XML processing on the source document before the transformation to RDF, and write a caveat in the specification recommending XPROC when this is the case."
DanC: the decision I voted for is to not mandate it
<chimezie> I was under the distinct impression we resolved the issue
<chimezie> it's just not testable
HarryH: what's everyone's
... we resolved to *not* mandate XInclude
... what should our test suite do??
DanC: shows this isn't resolved
Chime: difference between test suite and spec
<DanC> "If an information resource IR is represented by an XML document whose root node is linked to a GRDDL transformation TX"
(scribe is a bit lost in this discussion, trying to summarize shortly.)
MurrayM: thought we agreed that GRDDL works with infosets
<chimezie> 1) fetch 2) parse 3) handed over to GRDDL as XPath nodes / infoset
<chimezie> we are talking about step 2)
DanC: XPaths, specifically.
... angle bracket doc determines the infoset you get.
Chime: not my understand of what we resolved: 3 step process (as above)
DanC: I asked about changing spec, no one spoke up
MurrayM: agree that we didn't close the issue
HarryH: I thought we were clear
that we didn't mandate XInclude
... we wanted Murray to write an informative paragraph
Chime: what's our criteria for considering the issue closed?
DanC: record says we resolved it, I suggest we reopen it
Chime: I don't see the issue with not having a test case scenario for something we don't mandate
MurrayM: it's a function of having the right tools available
<HarryH> \me Harry is listening to the discussion to determine whether or not to re-open
<DanC> (you're actually allowing substantive discussion of the issue, which is out of order unless it's open.)
Chime: very distinct 3 steps
MurrayM: step 2 is ambiguous
Chime: what's our criteria for resolving step 2?
<HarryH> The question is whether if it's a test-suite implementation issue or a substantive issue. Apparently due to DanC's disagreement we should re-open the issue.
MurrayM: if we remain silent on XInclude and PSVI, then both approaches would be correct.
<HarryH> RESOLVED: re-open issue "faithful-infoset" as given by: (split it out of issue-mt-ns) and close it a la: The GRDDL WG does not mandate XInclude and further XML processing on the source document before the transformation to RDF, and write a caveat in the specification recommending XPROC when this is the case. (benA and BrianS abstaining)
HarryH: two approaches: DanC says we don't do PSVI or Xinclude, Chime says we don't express an opinion
MurrayM: is that DanC's opinion?
DanC: is this test suite or spec?
HarryH: this is a substantive discussion
RESOLVE to table the issue
DanC: Chime can't commit to
... I've copied test stuff into a different source control.
Chime: working on HL7 to RDF use
... not committed yet
... committing now where I can
... in Mercurial
<DanC> hg clone http://homer.w3.org:8123/ my-work-space
DanC: Brian has mailed a couple of tests
HarryH: we should get him CVS access
<chimezie> chimezie@Zion:~/devel/grddl-hg-test$ hg push http://homer.w3.org:8123/
<chimezie> pushing to http://homer.w3.org:8123/
<chimezie> ** unknown exception encountered, details follow
<chimezie> ** report bug details to email@example.com
DanC: we're not quite ready to approve the test cases.
<chimezie> File "/usr/lib/python2.4/site-packages/mercurial/localrepo.py", line 893, in push
<chimezie> lock = remote.lock()
<chimezie> AttributeError: httprepository instance has no attribute 'lock'
<HarryH> Propose we play test suite stuff meeting proper.
HarryH: I'll work on base-param
<HarryH> ACTION: To test #base-param and e-mail group [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action01]
<HarryH> Briansuda - test suite access?
<scribe> ACTION: DanC to give Brian McBride CVS access [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action02]
<scribe> ACTION: Ben to find access to GRDDL w3 cvs access [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action03]
<DanC> (well, benadida, collaborating by hg is a possibility)
HarryH: DanC, what do you recommend for source control? Hg, CVS, a combo?
HarryH: fastest path to last call
Chime: spec editing is the high priority
HarryH: for now, email Harry test cases.
<chimezie> nope, I didn't - i'll send it again
<DanC> ACTION: DanC to add N3/turtle mime type to Atom/turtle test case, noting the unregistered status [CONTINUES] [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action04]
<DanC> ACTION: DanC to write rules about XSLT 1.0 processing context [CONTINUES] [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action05]
<DanC> hGRDDL example
<chimezie> I don't have enough context for this scenario
<DanC> <head profile="http://www.w3.org/2006/03/hcard">
discussion of RDFa output use case (hGRDDL)
DanC: run all transforms, and chooses what to do with outputs
BenA: worried about how "clean" this approach is.
<chimezie> i'd like to try Ben's RDFa parser against the output of a clientside firefox transformation of this hGRDDL example
HarryH: might be worth building
... is it difficult to make test suite accept multiple correct outputs
DanC: there are cases in the spec
with multiple correct results
... either result is correct, as is their merge. So far test suite only has maximal result.
Chime: you'd have to explicitly do a non-lean-graph merging to handle this situation properly.
DanC: how about checking in one
graph, two serializations of it.
... even in the turtle case, I thought we would have only RDF/XML in the test suite.
HarryH: we won't get away with this in the RDFa case.
DanC: when you run the test, transformation produces Turtle, then compare graphs.
DanC: for testing purposes, the implementation of the hGRDDL test could produce NTriples
<HarryH> Chime, we'll move to your health-care primer test-case here.
<HarryH> After this discussion.
<DanC> . ACTION benadida: work on hGRDDL test, checking it in, and propose it to the WG for approval
Chime: test case with sample HL7
document (official HL7 example), associated with transform that
produces RDF using ontology I've been working on.
... working on whether stylesheet captures only sample doc elements, or larger portion of HL7 schema
... test case by end of this week
<HarryH> BrianSuda - primer?
<briansuda> still working on it
<briansuda> clearing up the sparql
<HarryH> All the same data is checked in.
<DanC> http://www.w3.org/2001/sw/grddl-wg/doc29/primer.html Revision 1.25 2007/01/09 23:54:10
HarryH: progress on hotel data section of primer, but not done
<HarryH> The word "Scraping" has been removed.
HarryH: word "scraping" has been
... issue of tag-soup HTML. Fabian is on vacation until next week.
<DanC> "Use case #8 - Pulling Data from the Web: Steffen wants to build a directory of the people he works with.
HarryH: removed word "scraping".
HarryH: Norm Gray: glad that it's
a use case to use tidy. But running a few GRDDL transforms to
try it out might be interesting as a separate use case.
... not sure we should have these kinds of use cases.
<DanC> "The script also systematically calls some classic transformations on the document in case these were not explicitly referenced in the page (e.g. FOAF and Dublin Core extraction, and the like)."
<HarryH> The script also systematically calls some classic transformations on the document in case these were not explicitly referenced in the page (e.g. FOAF and Dublin Core extraction, and the like).
HarryH: I could see that, but that's getting far from our goals.
RESOLVE to remove this sentence
<HarryH> ACTION: HarryH to remove the sentence: "The script also systematically calls some classic transformations on the document in case these were not explicitly referenced in the page (e.g. FOAF and Dublin Core extraction, and the like)." [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action06]
HarryH: haven't heard from Ian in a bit.
HarryH: faithful infoset caveat
MurrayM: haven't written it yet. started a good discussion, then dead over holiday, it would be good to pick up now.
DanC: waiting for spec text.
<DanC> ACTION: Murry to draft paragraph giving us caveat for faithful infoset issue closure. [CONTINUES] [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action07]
<HarryH> BrianSuda - do you want to write the content negotiation test-case?
HarryH: content negotiation use case is quite important
DanC: consider our weekly agenda
served in HTML or RDF.
... the online GRDDL service gets RDF and is then all confused
DanC: the spec should say "if you get RDF, you win! stop!"
<DanC> "Note issue issue-mt-ns is open. perhaps: special case for the RDF/XML namespace: RDF/XML documents are associated with RDF graphs as per the RDF/XML specification."
DanC: I have to think about this some more, it's on the todo list
<scribe> ACTION: DanC to write spec text for content negotiation use case. (Specifically recognizing RDF) [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action08]
<scribe> ACTION: HarryH to check in a test case on content negotiation [recorded in http://www.w3.org/2007/01/10-grddl-wg-minutes.html#action09]
HarryH: is everyone in agreement with DanC's explanation that, if you get RDF/XML, you stop?
Murray: it's not a GRDDL result
DanC: I think it is
... a degenerate case of GRDDL
... a GRDDL result is an RDF graph
... if RDF/XML comes back, then the work is done
Murray: there is no GRDDL
transformation, there can be no GRDDL result
... the goal of GRDDL is to get an RDF result from something that isn't
DanC: not necessarily
Murray: the conclusion that "it's a graph, so it's a result" is a leap
DanC: some parts of the spec won't work if we do it this way
HarryH: what parts?
DanC: P3Q case is problematic
<DanC> 3. Using GRDDL with XML Namespace Documents http://www.w3.org/2004/01/rdxh/spec
Murray: you're saying a GRDDL result is the same thing as an RDF graph
DanC: different vision of what the spec is supposed to do.
BenA: we should be careful to consider other RDF serializations, too.
HarryH: put remaining open issues in front or rear of agenda?
DanC: open first
<HarryH> Open issues first.
<HarryH> Meeting adjourned
<briansuda> HarryH, are you sticking around?
<HarryH> I have to grab lunch but I'll be online all day and am happy to play with SPARQL.
<HarryH> And work on primer stuff.
<briansuda> HarryH, i have just about everything working now, there is an issue with grokXFN.xsl
<briansuda> it doesn't seem to produce valid RDF
<briansuda> i'll email you have i have and finished so far