Erik Bruchez, Orbeon
John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Nic van den Bleeken, Inventive Designers
Roger Pérez, SATEC
Doug Scheppers, W3C
Steven Pemberton, CWI/W3C
Keith Wells, IBM
John Boyer: Nick, thank you for agreeing to chair the next two weeks.
Steven Pemberton: I hope to finish
this tomorrow.
John Boyer: I'll get the content in,
and the SOAP encoding and charset when I get back from vacation.
Then these are all the agreed changes, modulo implementation report
progress. Any progress?
Erik Bruchez: No progress,
unfortunately.
Keith Wells: I have a report on FF2
and FF3, with a few changes still to make. Where do I send
it?
John Boyer: The forms list is a good
start; we'll put it somewhere more permanent, perhaps a 2008
directory of reference documents.
Keith Wells: We can use XForms to
display the results as well.
John Boyer: It would be nice to have a
comparison of implementations and a summary column green/red for
each feature to show if we have enough.
Steven Pemberton: I am hoping for
some help on the response from Norm Walsh. He agreed to the
character clarification. Some people said this change reflected
reality in the RFC2396 to RFC3986. We thought the characters in the
URI had changed so you had to rev the language if it uses the new
base. He says the changes are editorial, as the range is entirely
what's allowed in XML. So, I think I am confused. I believe the RFC
reference change is real, but he says they only updated the
references section. If we agree I suppose we should just ask what
the RFC change means.
John Boyer: He does say some
characters need to be escaped.
Steven Pemberton: That's the normative
change, the semantic change that you need to convert the characters
to old to send them on the wire.
John Boyer: Has the Schema validation
rule changed?
Leigh Klotz: Aren't there some
weasel-words in there?
John Boyer: It's allowed to have stuff
that has to be escaped in a real URI. I think it says RFC2396
escaping, and then check to see if it's legal according to RFC2396.
So now XML base can provide a URI base that I don't know how to
process by RFC2396.
John Boyer:
http://www.w3.org/TR/2004/REC-xmlschema-2-20041028/datatypes.html#anyURI
So my processor will take the results of the (possibly relative)
URI and process it. It's conceivable it's expecting a certain bag
of characters. The bag of characters is changing with XML Base. The
processor for the resolved URIs has to understand that a different
set of characters is available. Even my own host language URIs have
to understand an updated set of characters relative to a set of
definitions from XML Schema part 2. Even if I don't, the upgrade to
base causes problems for the lower level processor, so the two need
to be updated at the same time (otherwise only XML base can use the
extra characters and the URI resolver would choke at some
times).
Steven Pemberton: I'll try to write a
reply that so we can agree on the issue.
Action 2009-07-9.1: Steven Pemberton to reply to http://lists.w3.org/Archives/Public/public-forms/2008Jul/0011.html to seek agreement on whether RFC2396/RFC3986 change is an issue.
John Boyer: What is the next point
about URIs?
Steven Pemberton: For href, the type
is anyURI, but in RDFa, rel specifies a thing where the algorithm
produces a URI, so rel can be seen as a URI; so, is it OK from the
XML Base point of view to apply that? Do they allow that? We
already use QNames in XForms and I think we've agreed that should
change to CURIEs in the future; so does XML Base permit you to use
things that aren't strictly-speaking URIs but can be
appropriate?
John Boyer: I thought perhaps they
would be absolute URIs, but I guess CURIEs can somehow be
relative.
Steven Pemberton: The namespace spec
doesn't require the prefix to be absolute.
John Boyer: So does this mean that XML
Base processing applies to namespace processing?
Leigh Klotz: I thought namespaces were
URNs instead of URIs?
John Boyer: The namespace spec says
"relative URI references is deprecated" http://www.w3.org/TR/REC-xml-names/
So switching from QNames to CURIEs would introduce the relative URI
problem to our documents. I know there was a schism and people were
using relative URIs in namespaces to point to Schemas, but they
stopped that. http://www.w3.org/TR/REC-xml-names/
Steven Pemberton: Do people agree
with me that the RFC change is a change in the syntax of the URI
reference?
Leigh Klotz: I think the sequence of
allowed characters in the XML serialization is the same, but the
meaning is now richer, which puts an onus on processors. with "The
range of characters allowed in xml:base attributes has always been
what are allowed in LEIRI in XML," Norm may be focused on the
document serialization and not on the responsibilities of the
processors operating on the document.
Steven Pemberton: Good comment.
Steven Pemberton: We are hoping
Mark Birbeck will look into this for XHTML2, but someone else
should from here.
John Boyer: Looking at the data types
would be good; we had some last-call comments on XForms 1.1 that
said we were doing things similar to XML Schema 1.1 so we should
put effort into finding those.
Latest http://lists.w3.org/Archives/Public/public-forms-tf/2008Jun/0000.html
Keith Wells: No response yet.
John Boyer: Maybe we should write off
to the HTML WG and ask if anyone wants to continue the conversation
on this document.
Action 2009-07-9.2: Steven Pemberton and John Boyer to look at HTML WG charter for schedule of delivery of last-call for Forms component of HTML and determine if escalation to HCG is necessary for better collaboration.
John Boyer: Nick can discuss in
your upcoming calls when you are chair.
Leigh Klotz: And Nick made a lot of
progress on the RNG schema for XForms 1.1.
John Boyer: We're working on a lot of
specs here, and we're concerned about there being too many. Some
other modularization specs tend to offer more than one module per
spec. So, does it sound reasonable to pursue, as the first public
WD draft model, to produce one spec with several modules?
Steven Pemberton: I'm personally
completely happy with one spec with lots of modules. I think Mark
would like to see separate specifications on the grounds that that
has been successful for RDFa, but personally I'm happy with one or
a handful of specs. Any other comments?
Leigh Klotz: Perhaps it can be
modularized here, published as one document, and then profiled with
multiple simple documents in the Backplane IG.
John Boyer: Yes, that sounds good.
Charlie's demo last week showed a submission module and the
instance element (no manipulators setvalue/insert/delete).
John Boyer: So we need to discuss this
further with Mark. Do you need to say something in the conformance
section that says you can adopt any of these?
John Boyer: What's an example of a
modular spec with a conformance section?
Steven Pemberton: XHTML
modularization.
John Boyer: So the wind is blowing in
the direction of bundling into the specs that are reflective of the
toplevel bullet points in the Future Features Wiki and the
sub-bullets are modules.
Erik Bruchez: [leaves]
John Boyer: Charlie has pulled
together the things we've talked about being in the instance
module. Mark expressed an opinion that the instance element could
be in a module by itself. But what does it really offer? It gives
an id attribute to a blob of XML data, and possibly src and
resource.
Leigh Klotz: And then instance
function idl.
John Boyer: That's a good point but
it's not clear to me yet...oh I see.
Leigh Klotz: That gives you the
formalism for getting the data out.
John Boyer: Submission could also have
access which refers to instances without XPath.
Leigh Klotz: It's the IDL function,
not the XPath function.
John Boyer: Or an IDL module might add
it.
Leigh Klotz: The IDL is generic so
it's not its own spec, like a JavaScript binding.
John Boyer: We could have IDL defined
for setvalue, insert, delete, etc. A subject for further
discussion.