See also: IRC log
Ben: we should target to complete our work by 1 Feb
Steven: what happens if we don't?
Ben: we make some sort of recommendation on where to continue the work, perhaps within HTML WG
Ralph: very clear sense from WG participants that having a point-of-contact within SW WG to this task force is important. The SWBPD charter may get extended a few months, should this work be bundled with that? There may be a new SW working group with new chairs, and this TF could attach there, too.
Ben: Guus and David made it clear they want output from this TF before 31 Jan
Ralph: we also get that pressure from others outside the WG
Steven: the HTML WG agreed at its f2f to go to Last Call. We have answered all comments; we are now replying to comments and doing edits to bring the document into Last Call shape. We expect perhaps a 6-week Last Call. There's a chance that there might be a second Last Call
Ralph: where does RDF/A syntax document fit in?
Steven: HTML WG would pull the necessary parts of the RDF/A syntax draft into XHTML2 WD simply due to maturity level dependencies
Ben: what about the RDF/A primer? would that be a separate document?
Steven: yes, we haven't included primer material in the XHTML specification document; it would be a separate document. The primer could easily be a Note
Mark: I would like to see CURIEs incorporated into RDF/A and thence into XHTML2
Ralph: CURIE needs to be REC-track to have any real benefit and if RDF/A syntax depends on CURIE then it should not be a separate document from XHTML2 spec
Steven: HTML WG has already accepted the CURIE idea in principle
Mark: the biggest issue came up when IPTC suggested they might not use namespaces as the binding mechanism
Steven: an interesting conclusion from the recent mail discussion is that many of the objections go away if we change the character from ':' to something else
Mark: responses to my blog point out that CURIE
is a way of codifying some existing practice in WiKis and elsewhere
... I point out that CURIE also tidies up QName usage; QNames can be reserved
for what they were intended for in XML
Ralph: I view Norm as a friendly reviewer -- if we are able to persuade him then we've likely done all our homework
Mark: Norm's biggest objection was that there might be two meanings to a given abc:def pattern; one interpreted using namespaces and one not. But I point out that this ambiguity already exists and in practice is resolved in context. In RDF/XML, the elements are interpreted as URIs
Ralph: and we're proud of that!
Mark: Norm seemed to be happier if there was no ambiguity that a CURIE might be a namespace reference
<Zakim> benadida, you wanted to ask about QName resolution
Ben: one course of action would be to review all our prior discussions and document what solutions we considered
Mark: CURIE work originated with IPTC requirement but since then we've recognized an issue with use of QNames -- that QNames are not an abbreviation for a URI. So the rationale for CURIE no longer depends on the IPTC use case; we have a stronger motivation. The TAG finding encourages people not to use QNames in this way unless they really have to
Ralph: RDF relies on QNames so this question may come back to haunt us
Mark: it's really RDF/XML that depends on QNames -- it's crazy that N3 syntax has to use them with their restrictions
<Zakim> RalphS, you wanted to ask about ':' vs something else
Ben: the argument for a change from ':' is to not appear to conflict with QName syntax but is there a real conflict? My impression is that there is not
Mark: the conflict may be more political
Ben: if we agree to change CURIE syntax, will we be forced to change other uses of QNames?
Ralph: I see an architectural compromise of the form "QNames SHOULD NOT be used as abbreviations for URIs" -- not MUST NOT but a possible future consideration for any abbreviated URI syntax could be whether it can be used as an XML element name
Steven: I don't see that reverse transformation as being necessary. Norm points out that an element name is a pair (prefix, localname) -- not a concatenation of two string
Mark: the issue may only be that RDF/XML uses the term "QName" too often. I described this in mail: "RE: CURIEs vs. QNames". I believe that CURIE can continue to use ':' without ambiguity, just as is done in XPath
Ralph: so if there is no actual ambiguity, I would argue that it increases the learning curve for users to have different syntaxes for QName and Abbreviated URI
Ben: if there is no technical conflict, I prefer to keep the ':'
Steven: there is a level of conflict that we resolve with square brackets so if we keep ':' we still have to do something else for, e.g. unadorned CURIE in href. We still need a syntax to distinguish CURIE and URIs
Ben: yes, the context will let us distinguish QName and CURIE
<Steven> SO if we use [dc:licence] to distinguish, then we could use [dc]licence instead (for instance)
Mark: how about dc[license]
Ralph: that looks like a URI; I could imagine a current use of href="dc[license]"
Mark: '[' is a disallowed URI character
Ralph: are you sure?
ACTION: Mark investigate authoritative specifications for '[' as a URI character [recorded in http://www.w3.org/2005/11/15-swbp-minutes.html#action01]
Mark: note that use of ':' in CURIE allows existing N3 documents to be correctly interpreted, as QName is a subset of CURIE
Ben: I am willing to work on the RDF/A primer, targetting an early December WG review
<MarkB> "The "national" and "punctuation" characters do not appear in any productions and therefore may not appear in URLs. "
<MarkB> from: http://www.w3.org/Addressing/URL/5_BNF.html
<MarkB> 'National' contains '[' and ']': http://www.w3.org/Addressing/URL/5_BNF.html#z57
Ben: we have several little issues remaining; src, role, etc.
<MarkB> (along with '^', '~', '{' and '}')
<MarkB> (...and '\')
Ben: regrets for 20 Dec
Ralph, Steven: regrets for 29 Nov
<Steven> It looks like [ ] are allowed: http://www.gbiv.com/protocols/uri/rfc/rfc3986.html#collected-abnf
<MarkB> no...just said they are not allowed.
<Steven> I think that that sytax has been obsoleted by RFC 3986
Ben: let's work on the smaller issues next week from the issues list
next meeting: 22 Nov, regrets from Steven
Change log:
$Log: 15-swbp-minutes.html,v $ Revision 1.2 2005/11/17 01:19:19 swick cleanup for publishing