W3C Forms teleconference June 18, 2008

* Present

Charlie Wiecha, IBM
John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Roger Perez, SATEC
Doug Scheppers, W3C
Steven Pemberton, CWI/W3C
Uli Lissé, DreamLabs
Keith Wells, IBM
Paul Butcher, x-port.net

* Agenda


* Previous minutes

http://lists.w3.org/Archives/Public/public-forms/2008Jun/0008.html IRC supplement: http://www.w3.org/2008/06/04-forms-minutes.html

* Modules

John Boyer: Let's talk about modules.

Nick van: I can't use much of the text there in the data binding modules because it constantly refers to other modules.
John Boyer: Which module?
Nick van: The bind module. The compute expressions are in the calculate module. The model isn't there yet. And in events, we now say when the events are dispatched, so for that I left text that says if binding expression fails. There is text about submission that needs to be moved. Is there a central place for that text to go?
John Boyer: The editorial note approach is the way to go. Then we can get the module out. Also you're adding the bind attribute to SNB and nodeset module. That module doesn't exist and has no author. http://www.w3.org/MarkUp/Forms/wiki/XForms_Future_Features . The bind module is part of the model bundle, and we might release them together. It says "Add @bind to SNB and NSB bindings" but if you look at the data model you will see "The Binding Attributes Module" contains them.
Nick van: So I'm doing too much work.
John Boyer: Also, instead of just a nodeset attribute, we talked about nodeset binding on the bind element.
Nick van: I wasn't sure where to put that. That's also the Binding Attributes Module?
John Boyer: In which case you inherit the thing. We need to find which of these modules come first in order to build them.
Leigh Klotz: We can use the topological sort algorithm!
John Boyer: If we had a graph. I'm pretty sure the instance data module is first, and binding attributes module is not far behind. Then other things can reference them. For example, the model module which Uli is taking on adds to attribute groups.
Leigh Klotz: Modularizing the attribute and element groups in the XForms 1.1 RNG I sent out in terms of the XForms Future Features list will help us find the plug-in points and dependencies between them at least in the syntax area; events, errors, etc. won't be covered here.

Action 2008-06-18.1: Leigh Klotz to modularize RNG Schema for XForms 1.1 and put it in the wiki.

Action 2008-06-18.2: Nick van den Bleeken to work on modularized RNG Schema for XForms 1.1 and help turn it into modularized XForms 1.2 schema for bind module and the binding attributes module.

Leigh Klotz: So is Nick doing Binding Attributes Module or is Charlie because Charlie is doing Instance Data Module?
Charlie Wiecha: It's OK.
John Boyer: The top-level bullets are more groups of specs that are going out at the same time.

Action 2008-06-18.3: John Boyer to write Binding Attributes module for XForms 1.2.

John Boyer: Uli is on the Model Module.

John Boyer: Charlie when can you be ready?
Charlie Wiecha: At least two weeks.
John Boyer: And you need the binding attributes module for setvalue, insert and delete.
Charlie Wiecha: I can get most of that done without it.

John Boyer: Then we have enough stuff to talk about next week; two weeks from now, we need Binding Attributes and Instance Data ready to talk about with the group.

John Boyer: Nick, in the meantime you're picking up the Relax from Leigh so you can do that before returning to bind.

* Upcoming Teleconferences

John Boyer: We need a summer questionnaire.
Steven Pemberton: [irc] http://www.w3.org/2002/09/wbs/32219/formssummer08/

* Next F2F (October 15-16 Virtual, Oct. 20-21 Cannes)


John Boyer: We have a virtual day, but we also have actual days. Do we have any special AV requirements?
Nick van: Maybe call in.
Charlie Wiecha: Maybe another try for webcast.
John Boyer: I think we'll have to do it ourselves. They mean LCD projector, flip charts, speakerphone. I think yes on all.

John Boyer: Are there groups we should be meeting? There's been an amalgamation of two working groups into WebApps. Do we need to present our modularization to them? And the JavaScript library?
Charlie Wiecha: You can also talk to the Backplane XG about doing a show-and-tell with a number of WGs.
John Boyer: I didn't see that there was an space for the Backplane XG members to meet.
Charlie Wiecha: We thought we'd schedule meetings with other working groups on Tuesday. We can talk about modularization points. The Backplane could show Dojo and Ubiquity integration, in terms of implementation.
John Boyer: So we should ask the WebApps group for time to present to them. Through Backplane?
Charlie Wiecha: I will do that anyway, but if you want a single request, we can do that.
John Boyer: In both cases it's modularization.
Charlie Wiecha: I'll take that action then.

John Boyer: Another group is the HTML group. Should we talk about streamlined syntax? Task force, Nick, Keith, Mark. Nick are you there?
Nick van: I'll ask. I think I can, for the whole week.

John Boyer: CDF is another. It may have fallen off our plate, but we still have mustUnderstand. Are we going to try to re-invigorate some logic? Leigh?
Leigh Klotz: I put in mustUnderstand to tell the XForms processor about required element and attribute extensions. We took it out because we decided it was also about telling the host processor, and that was a CDF responsibility. When we talked tentatively to the CDF they said they were only interested in namespace-free integrations and so it wasn't a good fit. We could try again.

John Boyer: There's also a new XML Security group. I'd like to talk to them.

John Boyer: We should prepare, not over the next two weeks, presentational materials for these groups. Charlie, would you (maybe through Backplane) produce (possibly with me) some material for the WebApps group?
Charlie Wiecha: That sounds like it makes sense.
John Boyer: Jointly?
Charlie Wiecha: Start, review, however.

Action 2008-06-18.4: Charlie Wiecha and John Boyer to create presentational materials for WebApps group for Tech Plenary.

John Boyer: Nick, can you provide materials for HTML WG?
Nick van: Yes, assuming I go.

Action 2008-06-18.5: Nick van Den Bleeken to develop create presentational materials for CDF group for Tech Plenary.

John Boyer: Leigh, can you help with CDF?

Action 2008-06-18.6: Leigh Klotz and John Boyer to develop create presentational materials for CDF group for Tech Plenary.

Action 2008-06-18.7: John Boyer and John Boyer to develop create presentational materials for XML SEC group for Tech Plenary.

* XForms 1.1 Implementation Report

Keith Wells: I am working on the Firefox 3 implementation report. Getting one that works is a problem, but then the report will be a couple of days.
John Boyer: Then others can follow your format. I can take it from there.

John Boyer: And for Chiba?
Nick van: I don't have time nor do Joern and Lars. They've never run the test suite.

John Boyer: Our current test suite is fairly manual.
Keith Wells: That's a fair statement. I've been working with Selenium, which is an open-source Apache-based license. We've been creating Selenium test cases. The hope is that that will be used as an automatic test suite. It doesn't handle open dialog, which is how Mozilla handles messages. It may require extensions for each implementation. I'd like to make it available.
John Boyer: I know Paul and Mark are interested as well. Is the correct path a way to change the tests not to use dialogs?
Leigh Klotz: It's a failure to follow our own MVC architecture so that supports changing them.
John Boyer: Anything you do to make it more automatic will pay off.
Keith Wells: It's a multi-stage target for us.
Leigh Klotz: Also, Mozilla's implementation of xf:message as s JavaScript alert isn't in agreement with what people are doing these days anyway, which is CSS for grayed-out displays.

* Action Item List


John Boyer: Mark has some done. Leigh as well.

John Boyer: Leigh, did you feel your work with ITS is done?
Leigh Klotz: I'm waiting for a document from Felix Sasaki. He wants us to publish it. I'll ping him.
John Boyer: Is ITS sufficient? What about ref?
Leigh Klotz: The label/@ref is already I18N. The referred document of labels is simply I18N'd itself.
John Boyer: So is this important for us?
Leigh Klotz: It's low effort. I'll ping Felix and ask if he has it done for us or if we need anything.

* SOAP encoding versus charset


Paul Butcher: In the spec, it mentions SOAP encodings. There are three possible ways to progress. One is to say that the encoding attribute doesn't have a result in the headers, another that it does unless there is a conflict in the SOAP definition (the best I think).
John Boyer: What's happening now?
Paul Butcher: In the spec or implementations?
John Boyer: The spec.
Paul Butcher: It says nothing; only data serializations. It doesn't say about headers. In the section about SOAP, there are quotes about charset MIME parameters.
John Boyer: So it's appended?
Paul Butcher: That's option A.
John Boyer: So what if an encoding attribute conflicts with the charset setting?
Paul Butcher: Yes, that's the issue.
John Boyer: <submission ... mediatype="application/soap+xml; charset=xyz" ... <John_Boyer> ... encoding="UTF-8"
Paul Butcher: Correct.
John Boyer: 1) what if encoding given and charset not given
John Boyer: 2) what if encoding given and charset also given but different
John Boyer: Is there a default if not given?
Paul Butcher: UTF-8
John Boyer: If the encoding is not given, then it's given. If the charset says ISO-8859-1 then it's a conflict.
Paul Butcher: Yes.
John Boyer: You said what happens now is the accept header is set. accept header has charset if and only if @mediatype expresses one
Paul Butcher: We don't exactly implement this in the message.
John Boyer: So option A has a "but" and you don't implement that.
Paul Butcher: Yes.
John Boyer: So option A is to keep current spec. option b is to keep current spec wording from above (accept header has charset if and only if mediatype expresses one). So your option A says if someone expresses a charset, use it.
Paul Butcher: That means it always overrides because it's UTF-8. I think that's wrong.
John Boyer: option b is to keep current spec wording from above (accept header has charset if and only if mediatype expresses one)
John Boyer: option A (preferable) is that encoding is used to set charset in accept header if charset is not expressed in submission mediatype
John Boyer: So the proposal is that in XForms 1.1 we switch to Option A.
John Boyer: proposal: In XForms 1.1 switch SOAP behavior to say that submission encoding provides charset to accept header if submission mediatype does not express charset
John Boyer: I suppose that will produce another test-suite test.
Keith Wells: I think so.
John Boyer: As we've made the changes to 1.1, are the test suites being updated.
Keith Wells: So far I'm doing OK.
John Boyer: As an example, required-but-empty back in validity.
Keith Wells: That one hasn't been made.

Action 2008-06-18.8: Keith Wells to add required-but-empty test back into validity.

John Boyer: Anyone see a problem with this proposal? No?

Resolution 2008-06-18.1: In XForms 1.1 switch SOAP behavior to say that submission encoding provides charset to accept header if submission mediatype does not express charset.

Action 2008-06-18.9: John Boyer to, in XForms 1.1, switch SOAP behavior to say that submission encoding provides charset to accept header if submission mediatype does not express charset.

* Meaning of expressed model(s) in form tag(s)


John Boyer: The link in the agenda the last time we talked about it.
John Boyer: Deferring to XForms 2.0 left this question: What happens to a model inside a form element?
Nick van: Didn't we say it would merge?
John Boyer: A separate model.
Nick van: No, a merge. The implicit UI model merges into the expressed model.
John Boyer: What if the form tag has two models?
Nick van: We decided it wasn't possible.
John Boyer: I think Erik said that was unnecessarily restrictive. Maybe just the first model gets the implicit behaviors merged?
Nick van: That seems reasonable, but if you have multiple models, you have to use the model attribute and then it isn't simplified syntax.
John Boyer: Definitely, but there aren't just two things (simplified and full).
Nick van: I think there's no problem; we had just decided not to allow it
Leigh Klotz: You're thinking about the problem of crossing the barrier.
John Boyer: Yes. I can't see a good reason to have more than one model in a form tag, but I'd be happy to say we pay attention to the first one, but Erik said it seemed restrictive for no reason.
Nick van: In HTML you can have two form tags.
John Boyer: Each can have its own model. UI controls within the form tag would not need a model attribute to talk to that model.
Nick van: If you add a separate model is that local to the children of the form tag or if you have two form elements is the first form element available in another?
John Boyer: We have that problem already once we have the model expressed.
Nick van: Since it has an id and you can refer to it.
John Boyer: That's the first level. If a person actually creates a model tag.
John Boyer: ...
Nick van: If you have no instance you can use lazy authoring. If you have an instance, ... If someone write the spec for the models then we can do it.
John Boyer: Somehow the action item landed on me for the streamlined syntax. There'll be a better lazy authoring story based on the name attribute. I can see the model expressed for submissions and instances for web services but the implied instance for UI. I'll try to do that.

* IRC Minutes


* Meeting Ends