W3C Forms teleconference February 18, 2009

* Present

Erik Bruchez, Orbeon
John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Paul Butcher, WebBackplane
Steven Pemberton, CWI/W3C
Charlie Wiecha, IBM

* Agenda


* Previous minutes

** Face to Face meeting at Google (Mountain View, California)


* Upcoming F2F

John Boyer: Steve Bratt has asked chairs to reduce the number of F2F meetings. We've already done that. Is everybody comfortable with the ones we have now?
Steven Pemberton: Yes.
Nick van: [irc] me too
John Boyer: Ok.

* Rich Web Application Backplane

Charlie Wiecha: Jack Janssen from CWI has implemented SMIL with Ubiquity. We're talking about SMIL code scheduling using the XForms "relevance" concept. It may be a unifying concept we use more broadly than XForms. Plus, the Ubiquity loader framework gives XML support in the client, as we discussed at TPAC and which I will discuss at AC.
Charlie Wiecha: I'm concerned we don't have an ODF rendering engine in the client. Things like numbered lists, recursion for bullets, can be a few classes (as a proof of concept) in Ubiquity.
Charlie Wiecha: Kevin Kelley has left IBM and has left the Backplane group. Please join us Tuesdays at 11.

Steven Pemberton: I think it's a good issue for W3C.
Charlie Wiecha: I think it's a good way of pointing out that the browser is not just an HTML engine.

* XForms RNC Schemas

Leigh Klotz: Given that we aren't pursuing the XForms 1.2 modularization as a top-level goal right now, making modularized RNC schemas for XForms 1.2 may not be as critical; however, breaking the files up into multiple sections (submission, form controls, etc.) is possibly useful.
John Boyer: Maybe as we get to understand new modularization ther will be.
Leigh Klotz: You can use the James Clark's NXML package from thaiopensource to get RNC validation of XHTML. If you want to get XHTML+XForms validation, use this http://xformstest.org/mode-20041004.tar.gz and do this

(push "/path/to/emacs/nxml-mode-20041004/" load-path)
(load-library "nxml-mode")
(setq auto-mode-alist
        (cons <code>("\\.\\(xslt\\|xhtml\\|xml\\|xsl\\|rng\\|xhtml\\|xsd\\)\\</code>" . nxml-mode)

Charlie Wiecha: I already have NXML installed.
Leigh Klotz: If you already have nxml-mode installed, just extract the schemas/klotz directory from the tar file and merge it in with your version. It adds the XHTML+XForms type to the XML/Set Schema/For Document Type menu to select XHTML+XForms.

* XForms output with no binding and no value


Erik Bruchez: I don't have strong feelings, but it doesn't seem there is a good use case for it.
Leigh Klotz: Do you feel the same about input?
Erik Bruchez: It's a mandatory single-node binding.
John Boyer: Yes, it's required. I think there has been some tendency to use label text using XForms rather than HTML content to make it easy to move content from one XForms supporting-type to another; XFDL vs. XHTML is my interest, but to some extent, ODF as well.
Leigh Klotz: But we don't prohibit it in the schema; I found this with the Mozilla implementation and couldn't prove it was wrong.
Erik Bruchez: Yes. I found the same.
Nick van: You can use a label inside a group.
John Boyer: Yes, label is not a form control.
Erik Bruchez: You can also put it in a case.
John Boyer: That's acceptable. But group is important because it says anywhere you have a label you can skin it in a form control by putting it in a group. But it is at least a form control. If I just want to output plain text, my options are quoted value attribute, or output with label.
Leigh Klotz: Doesn't output with label inherit the context?
John Boyer: It does but it doesn't it's not bound to anything. The text doesn't say. It says it can be used for two things but doesn't say what it does with neither.
Erik Bruchez: I know what Leigh is saying that the schema doesn't prohibit it but the spec does. If you don't have a containing form control, what does it do? It seems clear that it doesn't say what to do. I don't know how widespread it is. Our users don't because we don't allow it. I don't think there's much to do here, but I do want to clarify it.
John Boyer: We have had situations where we have edge cases on usages and we tend to try to make those work. That's my advocacy for making it work. Perhaps we should say that it shows the label.
Erik Bruchez: output doesn't allow inline text, though. That would make more sense. That has a parallel with HTML, which shows text if the element is not recognized.
Leigh Klotz: That's the same argument that leads to label being used.
Erik Bruchez: ...
John Boyer: Output with inline content with a value element would be an XForms 1.2 feature. If you have inline content rendered, and you also want to use label, what about the whitespace? And what about the inline content minus the label?
Erik Bruchez: The label would be output as a label element. But if it has a single-node binding... the only thing I am saying here is that it's not clear when there is no single-node binding and output itself for output. The label itself is clear.
John Boyer: For XForms 1.1, it's not going to be interoperable because we didn't say what to do if it has no single-node binding and no value. The easiest thing to do is if to fix the test cases not do it. In XForms 1.2, we can do something else.
Erik Bruchez: I agree; we don't need to fix it to solve the problem; just change the test cases.
Nick van: [irc] chiba doesn't allow it, I think
John Boyer: It's kind of late in the game for 1.1, and there's no compelling use case to make output do this.
John Boyer: So are people comfortable with the conclusion that we need to fix some test cases and not define the behavior?
John Boyer: Erik do you know which test cases they are?
Erik Bruchez: I modified a few and hit a few by chance, but I haven't looked.
John Boyer: Can you do this?
Erik Bruchez: I have the action of sending the modified version of the tests to Nick or John, or getting cvs access.

Action 2009-02-18.1: Erik Bruchez to produce modified version of the tests using output with no binding to Nick or John, or getting cvs access.

* XForms 1.1 Test suite changes


Nick van: Can I fix the appendix tests, by adding the ev:observer, that Keith was planning to do?
John Boyer: Test cases to modify: http://lists.w3.org/Archives/Public/public-forms/2009Feb/0035.html

Action 2009-02-18.2: Nick van den Bleeken to double-check modifications to tests to http://lists.w3.org/Archives/Public/public-forms/2009Feb/0035.html

Nick van: The tests have a listener for xforms-ready but they aren't in xf:model so we need ev:observer to attach to the model.
John Boyer: Couldn't you just move it to the model?
Nick van: It's easier to add the observer.
John Boyer: But it's a non-required feature so moving it would be better.
Nick van: OK, I'll undo the change and move them into the model.

Action 2009-02-18.3: Nick van den Bleeken to change Appendix B test to move the tests into the model.

John Boyer: Can you check these in and send mail to the list?
Nick van: OK.

* Ongoing test suite maintenance

http://lists.w3.org/Archives/Public/public-forms/2009Feb/0038.html http://lists.w3.org/Archives/Public/public-forms/2009Feb/0039.html

John Boyer: Can you periodically keep this list up to date? http://lists.w3.org/Archives/Public/public-forms/2009Feb/0049.html
John Boyer: There are some that need two implementations, and some need only one. We need this list to be updated when we get items in; we'll still have to double check when we think the list is empty, but keeping track from week-to-week would be helpful.
Nick van: I can do that also.
John Boyer: Thank you. There should be email to the list so we can see the archive. When we're ready to advance to PR, we'll provide full reports from Ubiquity. http://www.w3.org/MarkUp/Forms/2008/XForms11ImplReports/Chiba/FocusedResults.html
Nick van: I'll send in drafts now but later versions as we move forward another 5%.
John Boyer: The list is based on Firefox, EMC, Chiba, and Ubiquity, as far as I know. All are intending to provide full reports. Just operate off email for updating the running list. Don't produce an updated report; that will come from the implementors.
Nick van: I will create a wiki page then and use history.
John Boyer: That's an astoundingly good idea.

* Next Topics

John Boyer: What should we talk about for 1/2 hour?
Steven Pemberton: Which would you prefer?
John Boyer: I'd like 20 minutes on the XHTML for XForms assertions.
Steven Pemberton: If we wanted XForms 2, that would need more preparation work.

* XForms to XHTML Test Framework

Leigh Klotz: Can I suggest we just write the testable assertions, document what they should do, and let the Ubiquity test framework use Selenium to test its implementation? Let's avoid using XForms to write the success-fail validation framework. It works for XForms 1.0 and 1.1 but was difficult to get the dependencies right. For XForms for XHTML, let's just let the implementor (in this case Ubiquity) use its own test framework.
Nick van: Can we put the expected test results in an XML file?
Leigh Klotz: If there is a sufficiently abstract way to express it.
Charlie Wiecha: That's a good idea. Don't write the action handlers.

John Boyer: Sounds good.

** Intro Reading


John Boyer: Para 1 is must, shall, etc.
John Boyer: Motivation, attributes, script methods.
John Boyer: Local attributes. Other consuming host languages may use the XForms namespace.
John Boyer: So we just list local attributes for test cases; if someone wants to test an implementation with the XForms namespace, they would just say that in their implementation report.
John Boyer: The last paragraph says host processor should be namespace-aware.
Leigh Klotz: Why do we say that?
John Boyer: It's not required.
John Boyer: Host languages can adopt XForms elements.
John Boyer: I don't think there are any tests of conformance coming out of section 1.

** Form Containment


John Boyer: Each form is denoted by a form element and its content.
John Boyer: A form can contain a model; if it contains more than one model, all but the first must be ignored.
John Boyer: How do we collect the assertions together? In a wiki?
Leigh Klotz: Or in the spec; maybe generate them?
Charlie Wiecha: What tags in spec xml?
Leigh Klotz: I don't know.

John Boyer: We would have a minimal form with something minimal in it and someone would then decorate with detecting that this assertion has been met.
Charlie Wiecha: Yes.
John Boyer: If for every assertion we had test with markup in the text, wouldn't that bloat the text.
Charlie Wiecha: That's not what Leigh was suggesting.
Leigh Klotz: I think CSS printing and SOAP 1.2 for example just have anchors in the spec for each assertion.
Charlie Wiecha: Put RDFa on the paragraph.
John Boyer: I can put ids on the phrases but that doesn't make it clear to the reader.
Charlie Wiecha: Maybe in addition to diff mode you have assertion mode. A transform walks the XML and picks them up. Just an idea. We do have this RDFa technology. Steven? Does it seem reasonable?
Steven Pemberton: Sure.
John Boyer: So what might one write specifically to say this thing is an assertion.
Steven Pemberton: Work backwards from what you want to extract. Do you want the text of the thing? You need to say what the assertion is about; then you have a property that says it's an assertion; then you have the text of the assertion. You could use class or role, but if it's RDFa you have to say it's an assertion about something and what that something is.

Leigh Klotz: Take a look at http://www.w3.org/TR/wsdl20/#assertions and see that they have done design-for-test so that their assertions are called out with a dagger mark, and they highlight when you mouse over them.
John Boyer: And they have a table at the end.
Leigh Klotz: Probably generated with a transform like Charlie said.
John Boyer: I should explore this.

Action 2009-02-18.4: John Boyer to explore design-for-test assertions in XForms for XHTML using this as an example: http://www.w3.org/TR/wsdl20/#assertions

* Test Suite Encoding

John Boyer: What about these tests that say encoding="ASCII"
Leigh Klotz: They're wrong. I fixed the ones I had to edit.
John Boyer: Isn't it a no-op to change?
Leigh Klotz: Yes. It's just the checking in.
Nick van: I'll do it.

Action 2009-02-18.5: Nick van den Bleeken to fix XML encodings to use UTF-8 instead of ASCII.

* IRC Log


* Meeting Ends