See also: IRC log
RESOLVED: http://lists.w3.org/Archives/Public/public-grddl-wg/2006Nov/att-0066/grddl-wg-minutes-2006-11-15.html is a true record of Nov 15th meeeting
About the tutorial
Harry: currently talking with ben about join tutorial with RDFa. Waiting for initiative from BenA
ACTION: Harry to send info about WWW2007 GRDDL tutorial [DONE] [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action01]
Brian: is ok with comments on cross-intro
Fabien: not read yet
DanC: I would like wording to be suggested.
ACTION: Danja to suggest wording changes in light of recent comment cross-document introduction [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action02]
Danny: didn't have time yet to work on it but should be done by christmas
<DanC> test materials are http://www.w3.org/2001/sw/grddl-wg/td/xmlWithGrddlAttribute.xml v 1.1 2006/11/01 07:44:09 and http://www.w3.org/2001/sw/grddl-wg/td/xmlWithGrddlAttribute-output.rdf
Brian: any implementation?
DanC: yes 3 of them. every
implementation I know have run this test case.
... this is not a namespace test case.
Chimezie: in a lot of cases you spend time dereferencing namespaces you don't really need.
RESOLVED: http://www.w3.org/2001/sw/grddl-wg/td/testlist1#xmlWithGrddlAttribute is a correct and proper test case.
DanC: test on RDF from spreadsheet tools http://www.w3.org/2001/sw/grddl-wg/td/testlist1#projectsSpreadsheet
Chimezie: I haven't run this one.
DanC: I ran all of the 5 test cases
... we could wait for other testers to run this test
ACTION: Chimezie to verify http://www.w3.org/2001/sw/grddl-wg/td/testlist1#projectsSpreadsheet [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action03]
RESOLVED: http://www.w3.org/2001/sw/grddl-wg/td/testlist1#projectsSpreadsheet is a correct and proper test case .
ACTION: DanC to update test manifest to show these 2 tests approved [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action04]
Brian: "information resource" - kinda weird. Stephen King didn't write the page.
<DanC> (the page is a representation of something stephen king wrote, IMO)
ACTION: Brian to have a closer look at the Stephen King example in the spec [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action05]
Harry: I want to discuss Normans Gray's issue "error behaviour: The GRDDL spec doesn't say what a GRDDL processor should do if fed something which isn't one of these media types [text/xml, application/xml, and */*+xml], or which purports to be but isn't, or isn't well-formed."
DanC: the spec is written in terms of XPath so (...)
Danja: there are many pages out there using microformats so how should we handle them? use tidy?
FabienG: the question is non-wellformed XML but could still be used in a GRDDL transformation? JTidy being used in our wiki project but sometimes cleaning up HTML can destroy the metadata.
<danja> (I agree with the notion that HTML DOM -> GRDDL could work equivalent to XSLT, but also see Tidy as being problematic, no single algorithm)
<bwm> (me wonders what effect brackets round text has)
<danja> (stage whisper?)
HarryH: we only licence the use of GRDDL for XHTML and other use are allowed at your own risk.
<DanC> (comments like this are on the record, but sorta 2nd-class, since they weren't vocalized.)
ACTION: Fabien to add a sentence about non-XML HTML in the Use Case document. [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action06]
<bwm> ( i'm uncomfortable with "allowed at own risk" - GRDDL is defined on welformed XML and undefined otherwise. If someone transforms malformed xml into well formed xml and then runs a transform - that is up to them.)
ACTION: Harry to respond to Norman Gray [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action07]
Chimezie: it is not clear if you can access the right mime type within a stylesheet.
Harry: can we use the Atom test case
<chimezie> <xsl:output method = "xml" | "html" | "text" | qname-but-not-ncname media-type = string />
DanC: it's the job of the transformation to specify an RDF graph, by whatever means.
<danja> (Henry Story has an Atom to N3 XSLT 2.0)
<chimezie> We would be limited to resolving mime-types that are registered (RDF/XML is the only one)?
<danja> (me chuckles at facetious test cases)
DanC: the mime type we use should be registered mime types we could register a personnal mime type for N3
Chimezie: or use plain text mime type
DanC: or just leave mime type unspecified and output just plain N3 in text.
<DanC> (hmm... should the test materials have the N3 output, or the RDF/XML, or both? hmm.)
ACTION: DanJa to provide XSLT transformation from Atom to N3, and one case where it works [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action08]
ACTION: [WITHDRAWN] chimezie to develop test case to demonstrate both sides of the issue with #issue-output-formats [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action09]
<DanC> right; the 2 pending tests I have noted are: "no, xinclude doesn't come for free in xslt", "yes, you can use XProc to say to do xinclude and XSLT"
Chimezie: how to we deal with xinclude ? what about when xinclude does come for free?
DanC: when you point to a transformation it determines a relation between an infoset and an RDF Graph. When there is an XInclude it is ignored. The XSLT does not specify if we do XInclude first or not.
Murray: there is a policy question: if I am talking about the XML doc are we talking about the brakets or the infoset.
DanC: it is not about the angle brackets
Murray: is the content you have after you resolve the XInclude is part of the truth of the doc?
Murray: replace the XInclude by a triple stating there was an XInclude that wasn't resolved.
Chimezie: every transform would have to be aware of that.
<chimezie> the suggestion seems to be to add an additional test for an XSLT that will extract RDF which uses a term (such as rdfs:seeAlso) to link the source document with the XIncluded url
<chimezie> .. so the inclusion can be done within the RDF 'realm'
Chimezie: nothing in the XSLT specs mandates how the parsing should be done.
HarryH: how do we deal with layer processing in general?
<danja> (btw, possible useful ref for wording of "Normative Statement", XPath bit - http://www.w3.org/TR/xslt#section-Conditional-Processing )
HarryH: if we have two different answers from two different implementations we should show both and underline the problem.
DanC: I don't think the XPath doc should have processed the XInclude
Chimezie: this is an open issue.
<chimezie> <root xi:href="someDoc.xml"/> => <> rdfs:seeAlso <someDoc.xml>
Murray & Dan: give an example of a letter with three sections, the third one is an XInclude
<letter grdd:transformation="grokletter.xsl" >
<chimezie> though rdfs:seeAlso isn't expressive enough to capture what xinlclude is doing
Murray: in that document there is a GRDDL transform "grokletter" that knows how to handle this letter. grokletter.xsl has a pattern to match <xinclude> ...
<chimezie> so we have 3 test cases: 1) stand alone XInclude 2) XInclude mandated by an XML pipeline description 3) a transformation that 'describes' the inclusion using RDF terms
<chimezie> these 3 will be enough to talk through this issue
DanC: so the input still includes the XInclude.
Murray: yes so if the XInclude is
in the letter it will be processed by the transform
... if the XInclude has been processed before then your pattern won't fire. grokletter recognizes the presence of an XIclude and handles it.
Brian: I think the transform is the wrong place to deal with it - the publisher should get to choose whether to use an xinclude or not
chimezie> i agree with Brian. I think we can't resolve this without a policy decision
Brian: the transform should not depend on the implementation.
DanC: I could see both results being right.
Chimezie: the policy decision must be resloved before we resolve this.
<HarryH> ACTION: Murray to write up the pros and cons http://www.w3.org/2004/01/rdxh/spec#issue-mt-ns [recorded in http://www.w3.org/2006/11/22-grddl-wg-minutes.html#action13]
Harry: Ian and I achieved some things for the primer but still working on SPARQL
Harry: newt week scribe will be Danja.