W3C Forms teleconference January 13, 2010

* Present

Charlie Wiecha, IBM
John Boyer, IBM
Erik Bruchez, Orbeon
Steven Pemberton, CWI/W3C (Chair)
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Desginers
Uli Lissé, Dreamlabs

* Agenda


* Rechartering

Steven Pemberton: W3C management team are looking at our charter today. They will either give feedback, or assign reviewers today.

* XForms 1.1 Schema

* XForms 1.1 Test Suite Issues

** Tobi Krebs

http://lists.w3.org/Archives/Public/www-forms/2010Jan/0002.html http://lists.w3.org/Archives/Public/www-forms/2010Jan/0003.html http://lists.w3.org/Archives/Public/www-forms/2010Jan/0004.html

Leigh Klotz: I think output/mediatype/@ref is ok.
Nick van: He replied to your message and he is ok.

** Lars Windauer

http://lists.w3.org/Archives/Public/www-forms/2010Jan/0000.html http://lists.w3.org/Archives/Public/www-forms/2010Jan/0001.html

John Boyer: Action handlers in repeat are repeated. The one he's talking about uses ev:target to listen to the repeat element. So you get three action handlers, each responding to events dispatched to the repeat element. He says the test expects one, but that's not correct. We have another test with the action handler outside the repeat; it would be good to make sure we still have coverage for action handler inside repeat. So we should change the test to say we should see three of these.
Leigh Klotz: http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt10/10.5/10.5.a.xhtml
John Boyer: It says you should see an xforms-scroll-first with index set to 1.
Steven Pemberton: I can't see what difference it would make.
John Boyer: It says you should see one, but you see three. Which is what he should see. So he asks if the actions inside a repeat get repeated. These are inside groups inside repeat, so it's even clearer. The two possible fixes are (1) move action out of repeat or (2) change message to say "3 xforms-scroll-first messages."
Steven Pemberton: This test is testing the scroll-first event.
John Boyer: And scroll-last.
Steven Pemberton: The nesting isn't part of the test.
John Boyer: We get coverage from tests for things outside of their named feature, for example ev:observer.
Steven Pemberton: So the easiest fix is to change the label.
John Boyer: Right. 3 xforms-scroll-first and three xforms-scroll-last. We have action outside repeat elsewhere.
John Boyer: I vote going for the easiest change.
Nick van: You may have implementations that fail the test.
John Boyer: That's a challenge. Some may implement the event but not ev:observer.
Nick van: This is a combination of ev:observer and repeat.
John Boyer: Anybody who passes now will have most of that sophistication anyone.
Nick van: Those attaching only one action handler will now fail.
John Boyer: Right, but they're not following the spec.
Nick van: I know, but still.
Leigh Klotz: Now that we know it's at variance with the spec, we need to fix it somehow.
John Boyer: Chiba will now pass the test.
Nick van: Some implementations may fail.
Leigh Klotz: We'll have at least two correct implementations for this.

Resolution 2010-01-13.1: We change the label in test 10.5.a to say "3" instead of "2" messages.

** Joern Turner

Leigh Klotz: Are these done?
John Boyer: 5.2.1abc are done.

** 5.1.a


Nick van: I reported this at the F2F as well.
Charlie Wiecha: I think this one is done as well.
Leigh Klotz: OK so this one's done. That's all we need to test.

** 5.1.b

John Boyer: We didn't do anything.
Steven Pemberton: He's saying you can't get invalid for normalizedString.
John Boyer: Erik and I agreed that it's correct; you can't produce an xforms-invalid for this. Is there any content that could?
Steven Pemberton: He says not for normalizedString and there's no invalid character for string.
John Boyer: Yes, as long as it's well-formed XML.
John Boyer: Do we have the invalid case for string?
Steven Pemberton: No
Nick van: The text says everything but string.
John Boyer: That's in 5.1.a. So we should say "except normalizedString."

Resolution 2010-01-13.2: We change 5.1.b to add normalizedString as an exception to xforms-invalid test, as in 5.1.a (for string).

Nick van: He says it for token as well. I think space isn't allowed.
Leigh Klotz: I think it's normalized. It's NMTOKEN that doesn't allow internal spaces.
John Boyer: OK I'll fix it.
Leigh Klotz: I responded to Joern.

** 10.8.d

http://lists.w3.org/Archives/Public/public-forms/2010Jan/0000.html http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt10/10.8/10.8.d.xhtml

John Boyer: I think Ubiquity has an issue with this.
Nick van: Ah, the action is inside the instance data.
Nick van: Also the event is being dispatched to the wrong element as it goes to the parent. If you put the action in the model, the custom event should go to the model.
John Boyer: This is trying to test an event which bubbles. So we should dispatch the event to the xf:instance itself and then the action handler can be siblings of instance, because they will get the event as it bubbles to the model.
Steven Pemberton: What are we trying to do.
John Boyer: The label says "custom event that bubbles."
Steven Pemberton: We don't need an instance at all really.
John Boyer: We can attach the id="gohere" to the instance.
Leigh Klotz: The xf:instance element, not the instance.
John Boyer: Yes. Then we put the action as a sibling to xf:instance.
Steven Pemberton: So how does that match with "instance data is obtained by copying."
Leigh Klotz: That's just explaining why we can't have action handlers in the instance.
Steven Pemberton: Or put it in the group.
John Boyer: We only have to test that it bubbles. So put the listener as a model child. It will bubble from the xf:instance to the model.
Steven Pemberton: At the moment we produce two messages.
John Boyer: So we need a second handler.
Steven Pemberton: One on model and one on head.
John Boyer: They can both be in the model, but one needs to be ev:observer on xf:instance.
Steven Pemberton: That's good.
John Boyer: So ev:observer="gohere"
Steven Pemberton: So just move it out of the instance.
John Boyer: Right. I moved it and left one blank.

Resolution 2010-01-13.3: We target the event at xf:instance and move action handlers to a siblings of xf:instance.

John Boyer: I'll respond to Joern.
Steven Pemberton: And Paul Butcher.
John Boyer: I'll do that on Ubiquity.

* XHTML+XForms 11. RelaxNG Schema comments from ISO/IEC JTC1/SC34/WG1

http://lists.w3.org/Archives/Public/www-forms-editor/2009Dec/0007.html http://lists.w3.org/Archives/Public/public-forms/2010Jan/0010.html

Leigh Klotz: I need to go back on the ampersand issue and the idref issue.
John Boyer: Why are we calling them idrefs? The control attribute of setfocus for example. Maybe it's not an idref. A control can be repeated. The issue is that technically they are idrefs, but they are also ids.
Leigh Klotz: I think they are IDREFs; your form will fail.
John Boyer: No, one of our common attributes is ID. http://www.w3.org/TR/2009/REC-xforms-20091020/#structure-attrs-common

Action 2010-01-13.1: Leigh Klotz to investigate ID in XForms 1.1 RNC Schema http://www.w3.org/TR/2009/REC-xforms-20091020/#structure-attrs-common

John Boyer: What about the XML Schema?
Leigh Klotz: It would show up in the XHTML integration.
John Boyer: In common attributes it includes name="id" so everything that uses common attributes get id.

Action 2010-01-13.2: Leigh Klotz to investigate use of ampersand in RNC Schema.

Steven Pemberton: What are our plans?
Leigh Klotz: We can publish it in the zip now, or we can put it in the URL location again.
Steven Pemberton: Not in TR space.
John Boyer: We were going to create a separate space with elaborated directory structure. At some point we can move the link pointers in XForms 1.1 there, in errata. For some reason we put the XForms 1.1 schema in /2007 space, so when the RelaxNG schema came along I put it there as a Zip file. XForms 1.1 points to both of them in the /2007 directory. They're not in TR space and they're referenced by the spec.
Leigh Klotz: We can put it in /2007 or not, but not in a zip file.
John Boyer: /2007 is messy. So we should use a better structure.

Leigh Klotz: So we can publish the new ZIP at the current location, and we can publish the XForms for XHTML stuff at our convenience.

Action 2010-01-13.3: John Boyer to publish xforms11.zip from http://lists.w3.org/Archives/Public/public-forms/2010Jan/0011.html in current zip location.

Action 2010-01-13.4: John Boyer to publish files from xhtml+xforms11.zip from http://lists.w3.org/Archives/Public/public-forms/2010Jan/0011.html as individual files in TBD location for testing.

* Minimal content model for xf:group and text

http://lists.w3.org/Archives/Public/public-forms/2010Jan/0004.html http://lists.w3.org/Archives/Public/public-forms/2010Jan/0007.html http://lists.w3.org/Archives/Public/public-forms/2010Jan/0007.html http://lists.w3.org/Archives/Public/public-forms/2010Jan/0009.html

John Boyer: Test 4.3.6.a needs to be changed.
Leigh Klotz: John said that presentation of text other than labels and messages (such as in a group) outside of labels is an issue for the host language.
John Boyer: For example in http://lists.w3.org/Archives/Public/public-forms/2010Jan/0007.html to get ODF to include XForms form controls in the future, their text and form control layers are different layers.
Leigh Klotz: It's not all form controls, just group, case, and repeat.
John Boyer: I don't know how the ODF thing will get decided but it seems like not a good idea to make an addition to change the schema now from what the spec and XML Schema says. Our XML schema isn't optimal, but if you look at UI Content, it says that it can be a number of outputs; if you look at label, help, hint, alert, and message, they are complexType mixed="true". Those five specific elements can mix it. The spec is right and the schema is right. I don't know how to say if it's easier to say that the definition of UI.Content is where the mixed="true" is coming from. Maybe you can't do that.
Leigh Klotz: You can, but this is a question of what we want to do, not how.
John Boyer: I'd advocate that we not open up text content in container form controls.
Steven Pemberton: What do we do?
Charlie Wiecha: Fix the test case.
Leigh Klotz: Yes.
Steven Pemberton: 4.6.3.a
John Boyer: I'll fix it right now.

Resolution 2010-01-13.4: We fix 4.6.3.a to use group/label/text() instead of group/text().

* IRC Log


* Meeting Ends