W3C Forms teleconference February 2, 2011

* Present

Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Erik Bruchez, Orbeon
Steven Pemberton, CWI/W3C (chair)
Philip Fennell, MarkLogic [irc]
Kurt Cagle, Invited Expert
Dan McCreary, Invited Expert
Alain Couthures, AgenceXML

* Agenda


* F2F Meeting

Steven Pemberton: There is a TPAC at the end of the year. Do we want more than that?
Kurt Cagle: How many people are going to Balisage?
Steven Pemberton: I can't make August. Prague would have been nice.
Nick van: It's fully booked.
Leigh Klotz: TPAC is in Santa Clara again I think.
Steven Pemberton: So where and when for next F2F, or virtual?
Nick van: I'd like to have one in the first half of the year.
Steven Pemberton: Or in Amsterdam.
Nick van: Or the US this year.
Kurt Cagle: SemTech, in the Bay Area.
Erik Bruchez: June 5-9, downtown San Francisco.
Steven Pemberton: It's not necessarily an XForms-allied conference.
Kurt Cagle: Dan McCreary and I will be going, but it's not strongly affiliated.
Steven Pemberton: San Francisco is a good location, but we need a host.
Leigh Klotz: I can host and I'm sure Erik would be ahppy to have it here.
Steven Pemberton: Normally we meet 3 or 4 days.
Nick van: I can't leave home the end of July.
Steven Pemberton: So maybe the beginning of June, on the west coast.
Steven Pemberton: And do we need a one-day virtual meeting?
Erik Bruchez: That sounds fine. Progress is a little slow and we've made significant progress in a few hours.
Steven Pemberton: Others aren't here right now so let's pick just a tentative date for a one-day meeting. March 10th? Let's pencil that in.

* First Public Working Draft of XForms 1.2

Steven Pemberton: It's time to plan this. We're late on our tiemline. We could use the time from now to the one-day meeting.
Steven Pemberton: We ought to be able to do it earlier, but having the meeting to discuss remaining issues would be a good idea. With four F2F per year we used those as heartbeat.
Kurt Cagle: Then March release and June/July release.
Steven Pemberton: Then by beginning of March we should have a draft to publish after. Sounds good?
Kurt Cagle: Sounds good to me.
Steven Pemberton: Alright.

* Multi-lingual web


Steven Pemberton: I'm representing XForms for this meeting in Pisa.

* XBL2

http://lists.w3.org/Archives/Public/public-forms/2011Jan/0030.html http://lists.w3.org/Archives/Public/public-forms/2011Jan/0031.html

Steven Pemberton: Last week I said we'd raise it at the HCG. There will be a discussion there in 10 days time. Leigh, can you call?
Leigh Klotz: When is it?
Steven Pemberton: Friday week, an hour earlier than xml-cg.
Kurt Cagle: I'd like to join.
Steven Pemberton: It's member only, but we invite others. Kurt, you're more than welcome to join.

* XPath 2.0

Nick van: I've been working on XPath 2.0 module http://www.w3.org/MarkUp/Forms/specs/XForms1.2/modules/xpath20/index-all.html
Nick van: We need to discuss the XPath 2.0 evaluation context. It's now more well-defined which things are stable. I don't think we made a decision yet. http://www.w3.org/MarkUp/Forms/specs/XForms1.2/modules/xpath20/index-all.html#expr-dynamic-context

A number of functions specified in [Functions and Operators] are defined to be stableFO, meaning that if they are called twice during the same execution scopeFO, with the same arguments, then they return the same results...

Nick van: So we should define in XForms how our current date and time functions act: never, refresh, recalculate, rebuild, implementation-defined?
Steven Pemberton: I expect in my use cases for them to be updated between refreshes.
Nick van: Is it also in a bind calculate, if it refreshes, how do you find it there?
Steven Pemberton: So there's no "dirty" or impure function?
Nick van: We can define how our functions are called. If you have a calculate that uses the current date, no two implementations have the same behavior at the moment. rebuild, refresh?
Kurt Cagle: This brings up another issue, control related: it might be worth thinking about, on a given control, putting a timer that indicates the granularity of refresh. An output, for example, with a timer granularity of 1s to force 1s refresh. I know a couple of times I'd like to have that in time-dependent systems.
Steven Pemberton: We have something like that, on incremental.
Nick van: An output that displays the current time, in the spec it never changes its value. Is that the problem you're bringing up?
Kurt Cagle: Incremental only indicates when you have a change in the model. This is a change dependent on a function in a binding.
Steven Pemberton: True enough.
Nick van: If we say it's refresh time, then no refreshes are triggered. And output would never update.
Erik Bruchez: If you want a timer, you dispatch a delayed event and set a value. I think it's acceptable for a rare situation. Same in HTML4+JavaScript, with settimeout.
Nick van: The wiring you have to do yourself.
Steven Pemberton: Refresh sounds like the right time.
Nick van: How do you know to do recalculate? Or is only during rebuild?
Erik Bruchez: This isn't new with XPath 2.0.
Nick van: Now I'm looking at XSLT's definitions and there is a paragraph about initializing the dynamic context. So we know it's not interoperable, but we need to say something.
Steven Pemberton: That was one of the points, Nick. I must say I'm very impressed by the document, Nick. It's very good.
Nick van: For the rest, we dicsussed on the previous call. There are still things to do and dependencies on other XForms 1.2 work but I can go forward.
Steven Pemberton: Are there areas you want us not to miss?
Nick van: I copied the functions and changed the types. The first sections are the new ones: version attribute, initializing static/dynamic context (based on XSLT).
Steven Pemberton: This is a separate spec. The idea is that we produce this as a module separate from the rest of the XForms 1.2 spec?
Nick van: I'm not sure. I started this when we were doing modules. But I don't have a strong feeling. If it's not, I have to copy everything back into the wiki.
Steven Pemberton: Do we have a consensus about the way we want to do this?
Leigh Klotz: I think we don't have to decide now; this is like one of John Boyer's thin specs, even if we decide not to.
Steven Pemberton: So should we announce this to the list for comments?
Nick van: There are some broken links so it won't pass pubrules.
Steven Pemberton: It's publicly available though so we can announce it for discussion.

Kurt Cagle: I've been watching Michael's work with XSLT 2.1 into 3.0. There will be XPath 3.0 and XQuery 3.0. Do we need to look at this with regard to XForms? Could we harmonize the versions? The downside is that we'd need the XPath 3.0 implementation supported. Changes between 2 and 3 aren't that radical. Is it worth thinking of XForms 3.0?
Steven Pemberton: We're not chartered to to that.
Kurt Cagle: For the future.
Nick van: Wasn't the rationale because the XQuery / XSLT / XPath are all from the same WG, and they share the same function library? Do we want to be on the same release track?
Kurt Cagle: The downside is that it does put them on the schedule.
Steven Pemberton: In the past we've trailed them so we get solidified implementations. We didn't put XPath 2.0 in until we had implementations, so there is some advantage to trailing.
Erik Bruchez: We're very conservative with version numbers. I think XForms 1.1 was supposed to be quick and it took 6 years.

Steven Pemberton: I believe you're suggesting we should call it XForms 2.0 if we use XPath 2.0 and keep our version numbering in line.
Kurt Cagle: I think it's at least worth thinking about. I think 1.1-1.2-1.3 indicates XPath 1.0.
Steven Pemberton: It has a certain charm of argumentation. Erik, you were thinking of XForms 2?
Erik Bruchez: Yes, but I don't know about 3.
Steven Pemberton: Would anybody object?
Leigh Klotz: Is XPath 2 going to be required or optional?
Erik Bruchez: I think it's going to be optional.
Leigh Klotz: So we'd change our version number because of an optional feature?
Steven Pemberton: Eventually we would.
Leigh Klotz: So XForms 1.2 would have optional XPath 2.0 but XForms 2.0 would have required XPath 2.0.
Steven Pemberton: Who wants it to be optional?
Nick van: John and maybe Alain.
Alain Couthures: No, I think it is a good idea.
Erik Bruchez: There's been a lag, but we never supported XPath 2.0 in our implementation of XForms. It's a huge benefit.
Kurt Cagle: With XQuery in the browser, we're beginning to see more solid near or at XPath 2.0 implementations in the browser, plus Orbeon. The argument from the browser side is weak. It's the non-HTML implementations that might be more of an issue.
Steven Pemberton: I propose that we discuss this with John and point out we're all in favor of XPath 2.0 as required. OK?
Alain Couthures: OK

* W3C HTML/XML Task Force


Kurt Cagle: There hasn't been a lot of activity. There was a short meeting, possibly without a quorum. The activity has been mostly looking at the use cases. There is a use case #6 which is XForms related and I'm working on that and I'll send it here.
Steven Pemberton: What are the main issues being discussed?
Kurt Cagle: Use cases for combinations of XML and HTML and places where they force requirements on one or another, such as XML Data Islands in Use Case 4. Other cases are essentially XML embedding HTML, HTML embedding XML, script elements and bindings. There's been nothing definitive other than where use cases involve interactions between the two. The HTML side has been trying to minimize cases. Michael Kay may be less involved, but he's been writing XML support arguments. Another issue is where HTML ends and XHTML begins, and where you can't use XML within HTML. That's not been fully fleshed out, but it's Use Case #5. Those are in development, and we'll discuss those next week.
Leigh Klotz: And XBL touches on that as well.
Kurt Cagle: Absolutely. And at what point in the process can we deal with mixed embedded content. xf:repeat, etc.
Steven Pemberton: So we should draw their attention to this.
Kurt Cagle: The biggest issue is XBL namespace removal.

* XHTML+XForms

Leigh Klotz: I did RNC, and Owen Newnan did XSD and Philip did NVDL, and you suggested we look at XHTML Modularization. So we need to pick one.
Steven Pemberton: We need to define where our bits go in XHTML Modularization.
Leigh Klotz: We can't use DTD's.
Steven Pemberton: We can use XSD and Relax.
Leigh Klotz: We've got those both.
Steven Pemberton: So you'd like to talk with Shane?
Leigh Klotz: Yes.
Steven Pemberton: Let's make that a real action.

ACTION-1775 Steven Pemberton and Leigh Klotz to contact Shane McCarron about using M12N for HTML+XForms.

ACTION-1776 Steven Pemberton to invite Leigh and Kurt to HCG XBL meeting.

* IRC Minutes


* Meeting Ends