Alain Couthuries, AgenceXML [IRC only]
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Philip Fennell, Marklogic
Steven Pemberton, CWI/W3C
Erik Bruchez, Orbeon
Uli Lissé, DreamLabs
Kurt Cagle, XMLToday
Steven Pemberton: Next week the
calls in Europe will be an hour earlier, for two weeks.
Nick van: I won't be here next
week.
Philip Fennell: March 16th and
23rd.
Steven Pemberton: It's this Friday
again.
Leigh Klotz: I'll join. Claudius
Teodorescu has asked me about XBL and XSLT PI support in desktop
browsers again.
Steven Pemberton: I believe we
haven't done enough preparation. Should we find another day?
Kurt Cagle: We can do it after the HCG
meeting
Steven Pemberton: The week of the June
6-8 is not possible. Dan MacCreary and Kurt are going to the
Semantic conference. It's a holiday weekend for me.
Kurt Cagle: How about pushing it back
a week?
Steven Pemberton: May 23rd but the
reason for us coming has changed. The 13th is a holday.
Steven Pemberton: Nick, what's the
latest you can make it?
Nick van: I can't to travel in
June.
Steven Pemberton: THe TPAC is the week
of the 16th in Bilbao, in Spain, and I'll be tehre all week.
Nick van: Then you're travelling two
weeks in a row.
Steven Pemberton: Otherwise May 30th,
31st, and July 1st. But Dan and Kurt would travel twice.
Kurt Cagle: I'll be coming
anyway.
Leigh Klotz: May 30th is Memorial
Day.
Kurt Cagle: How about the days after
the conference? I'd prefer that for travel anyway, before or
after.
Nick van: When is the last day of the
conference?
Steven Pemberton: June 5th-9th, but
the next weekend is a European holiday weekend.
Nick van: That's not a big problem for
me.
Steven Pemberton: It's somewhat of a
problem for me.
Kurt Cagle: Memorial day weekend is
not a big issue for me.
Steven Pemberton: But it won't work
for Leigh.
Steven Pemberton: I'll make a
questionnaire.
ACTION-1780 Steven Pemberton to make questionnaire for early-summer/late-spring F2F dates.
Kurt Cagle: If we are looking at an
appearance attribute, we should consider what that means for full,
minimal, etc. We currently don't have those.
Steven Pemberton: I don't see the
message.
Leigh Klotz: It's called "Re:
Draft"
Leigh Klotz: Alain and I have done
something with an appearance attribute and an href so that in
XSLTForms you can get the label be a link that displays the
help.
Kurt Cagle: That's sweet.
Steven Pemberton: I see, it's kind
of a pain to get block and inline labels.
Kurt Cagle: I'm not sure of the
approach, but if we are talking about ancillary elements having
appearance attributes, we should figure out what that means and
give general guidance.
Steven Pemberton: What do you mean by
"minimal could be associated with a default label"?
Kurt Cagle: A textbox with the label
inside it in grey, which goes away.
Steven Pemberton: That's a rather nice
way to express that pattern, and it's trick to do right. That's
quite nice, actually. Having an easy way to do block and inline
labis is nice. Did we have an action to write up what we agreed
on?
Steven Pemberton: Kurt, would you
do the write up?
Kurt Cagle: Sure.
ACTION-1781 Kurt Cagle to write up the results of control labels and appearance as a wiki entry to use as a basis for discussion at virtual F2F day.
Steven Pemberton:
Steven Pemberton: They've published
the minutes as a public link.
http://lists.w3.org/Archives/Public/public-hypertext-cg/2011JanMar/0007.html
Steven Pemberton: Art Barstow, who is
chair of WebApps, will be there.
Kurt Cagle: Do we have a plan for what
the group wants to say?
Steven Pemberton: I think we agree
with what Leigh propounded: a superset version with namespaces
which adds namespaces and can "compile down" to WebBL. Leigh?
Leigh Klotz: I think we should ask
Erik.
Erik Bruchez: Yes, sound good.
Steven Pemberton: And two documents,
or two parts?
Erik Bruchez: I assume that whoever is
producing the spec would produce the second version.
Steven Pemberton: No, this works
around that.
Erik Bruchez: I thought Ian Hixson
removing namespaces was an experiment. Who is the editor?
Steven Pemberton: He's been editor all
along.
Erik Bruchez: Then whoever is working
on that part, the core, is not going to be from this group, and it
won't have namespaces. The only thing we can do is find areas that
don't allow extensions and provide feedback. We can't decide that
there's going to be a full spec done by someone else.
Leigh Klotz: The W3C team members
rejected the idea of optional features because of "patent policy"
issues. So the plan was to have a second profile spec which
includes the first spec and adds namespaces.
Steven Pemberton: I've never heard
about this patent policy issue before; I'll have to go ask some
experts.
Kurt Cagle: It did sound like one of a
series of red herrings.
Erik Bruchez: I don't think we can
have much influence on the people who are doing the work. If the
WebBL people make specs, we cannot control how that's going to be
done. I think we shouldn't waste too much time on it, and just
provide initial feedback.
Leigh Klotz: I think the current
proposal is to take "WebBL" (though it won't be called that) and
define a new spec called "XBL" (though it won't be called that)
whichs says "Take WebBL and add namespace syntax to it."
Erik Bruchez: That sounds OK.
Steven Pemberton: We can call it
whatever we want.
Kurt Cagle: I'd like to caution: right
now when you talk about HTML5 and XHTML5, the underlying message I
get is that it's just HTML5 written in XML. At the same time, what
we're seeing here is that there are differences (from namespaces,
from content, namespace and schematic bindings with data types for
entities within XHTML5 that aren't in HTML5). These things aren't
covered in the HTML5 specification. It may be that one thing we
have to do is to ask who is responsible who is responsible for the
XML implementation and how we handle issues where there are
pre-existing technologies that are bound to XHTML that are
irrelevant to HTML5? XBL is a perfect example; there's more to it
than just namespace bindings. There are other aspects.
Steven Pemberton: That's a good
argument to put forth on Friday; it's not enough just to say that
XHTML5 is HTML5 in XML clothing, becuase it does imply a lot more.
If XBL is going to support both of them, then by definition it
needs to support ...
Kurt Cagle: So we need to ask HCG who
the stakeholders are of XHTML5 and at what point are there going to
be divergences?
Steven Pemberton: OK, done?
Steven Pemberton: The call will be the
same as last time, an hour earlier than this call on Friday.
Kurt Cagle: One question I hear is
whether AVT should migrate into XForms itself: can you have {}
within an XForms attribute should as a value, or where we have
fixed bindings such as resource attribute. Or should we strictly
limit AVT to underlying host language, and keep it out of XForms
itself? Have we resolved that?
Erik Bruchez: There was no decision.
Value isn't a good example because it's an XPath expression.
Steven Pemberton: It's a good example
because we need to ask.
Erik Bruchez: It's not allowed.
Kurt Cagle: I agree, I mean more like
the fixed attributes.
Erik Bruchez: Where do we get AVTs
from? There are attributes in XForms that are static.
Steven Pemberton: An historical
mistake.
Erik Bruchez: Look at HTML4, there are
static attributes but they're not really static because of script.
An XForms implementation could do the same, but it's not
declarative. So we needed a solution in our implementation that was
simple and we used AVT from XSLT. It doesn't require JavaScript or
XQuery scripting.
Erik Bruchez: In XForms 1.1 we did an
incomplete effort to control such fixed attributes such as load,
and we documented those in the spec with nested elements, but
there's no general rule. It's not very satisfying, and the syntax
is absolutely horrible. AVT expresses this way, way shorter. We
should be trying to use shorter attribute names as well.
Leigh Klotz: And the second use, for
host languages?
Erik Bruchez: Yes, Kurt was asking
about on XForms elements, but yes we support them on HTML, SVG,
etc. That gets transmitted into an XForms extension element called
"control" with an idref to the host element. It is a lightweight
syntax to modify attributes in the document.
Steven Pemberton: I like the idea of
the generality. I fear the complication of the user having to
remember whether to put squiggly brackets or not. Maybe there's no
way out of it because of our history and you need to know that
anyway.
Nick van: There are a limited set of
those that are XPath: nodeset, ref. Are there others?
Erik Bruchez: It should be less to
write about in the spec. "For every attribute that is not an XPath
expression, you can modify it dynamically..." Erik you can say
"may/must" support nested elements. It could be one paragraph. It's
clearly a new thing for form authors, but the upshot is that it's
more flexible.
Leigh Klotz: You just said it was one
paragraph to write; is that an action?
Leigh Klotz: In addition to nodeset,
ref, and XPath values (such as MIPs) we shouldn't allow ID.
Erik Bruchez: Yes, it gets messy if
you allow that.
Leigh Klotz: What about idref, such as
submit/@submission or toggle/@case?
Erik Bruchez: For the actions there we
support the AVTs.
Leigh Klotz: So we disallow it on ID
but not on IDREF.
Erik Bruchez: Yes, setfocus, send,
submit etc. would work. There is a need for dynamism there.
Philip Fennell: There's one problem on
IDREF: you can't validate as IDREF anymore.
Erik Bruchez: You wouldn't. You can't
determine it until runtime. So having {} attributes that are
runtime poses concrete problems if you need to have strong
validation. It shouldn't be a showstopper.
Kurt Cagle: Are you validating against
the live instance?
Leigh Klotz: Just define the schema as
IDREF or at least one matched pair of {}'s.
Erik Bruchez: It's hard to validate
IDREF or {}'s. You can write a rule that matches and multiple {}'s
and nested XPath expressions, but the regexp language in XSD is
fairly weak. IDEs would be expected to do a better job.
Kurt Cagle: You don't know at design
time, but once you have the instance you can validate.
Leigh Klotz: We need a weak version of
design-time validation, so that we don't break editors at least. I
think the IDREF|{} works around it.
Philip Fennell: So we drop the bind
IDREF validation issues for NVDL?
Leigh Klotz: I've never been able to
get IDREF validation to work in RelaxNG either. I think it's a
design problem with IDREF.
Kurt Cagle: Is this an oXygen
problem?
Philip Fennell: No, the same problem
appears in XProc.
Erik Bruchez: We can't limit ourselves
to a schema becuase IDEs can do more.
Leigh Klotz: I think we can publish a
schema that works and allows IDREF or AVT.
Nick van: What about fixed
enumerations?
Leigh Klotz: You write
appearance={"full" | "minimal" | "compact" | AVT } and AVT =
text { pattern = ".*{.*}.*" } and then refine the definition
of AVT over time (as multiple XPath expressions) if you want. But
this at least makes it work at design time.
Steven Pemberton: Erik would you be
willing to write this up?
Erik Bruchez: I think I already have
an action to own AVT discussion.
Kurt Cagle: I wouldn't take one
paragraph as a hard limit.
ACTION-1782 Erik Bruchez to write one paragraph description of proposed spec text for AVT.