W3C Forms teleconference December 2, 2009

* Present

Charlie Wiecha, IBM
Erik Bruchez, Orbeon
John Boyer, IBM (chair)
Kenneth Sklander, Picoforms
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Steven Pemberton, CWI/W3C
Uli Lissé DreamLabs

* Agenda

http://lists.w3.org/Archives/Public/public-forms/2009Dec/0002.html

* Previous minutes

http://lists.w3.org/Archives/Public/public-forms/2009Nov/0024.html IRC supplement: http://www.w3.org/2009/11/25-forms-minutes.html

* Virtual F2F Day Thursday

John Boyer: Steven, I know you won't make that one. How many more before the break?
Charlie Wiecha: I'm here until the 23rd.
John Boyer: I'm here until the 16th.
Steven Pemberton: So the 6th will be the first call?
Nick van: [irc] can't make 30th december too
Charlie Wiecha: So we drop all three?
John Boyer: Do we want to drop the 6th?
Steven Pemberton: What's the problem with the 6th?
John Boyer: None for me. So we just drop December 23 and December 30?
Steven Pemberton: I'm good for the 6th.
Nick van: fine for me
John Boyer: OK, our last call will be December 16th, and the first will be January 6, 2010.

Resolution 2009-12-2.1: Our last teleconference for 2009 will be December 16th, and the first will be January 6, 2010.

* Rechartering

John Boyer: I spoke to Phillipe
Steven Pemberton: Phillipe has been away from work.
John Boyer: When do we start with new calls? The 6th?
Steven Pemberton: I'm happy to help with the transition.
John Boyer: Thank you. You'll chair on January 6th?
Steven Pemberton: We should get together the day before for the agenda, by email?
John Boyer: I have the running agenda document. I'll send it to you on the 4th.

* Update TheMatrix

http://www.w3.org/QA/TheMatrix

John Boyer: The matrix is updated for Rec. The validator checkbox should be there for XForms 1.1. Ian Jacobs asked for the link to the validator I gave him the schemas.
Steven Pemberton: W3C has an online schema validator so we should be able to make a form that uses it.
Leigh Klotz: I've also heard from the Relax ISO committe that they have found some mistakes in our RNG schema and will be sending us info.
John Boyer: Updates are welcome.

* Drafts provided by Owen Newnan

http://lists.w3.org/Archives/Public/www-forms-editor/2009Nov/0003.html

John Boyer: These combine XForms 1.1 and XHTML 1.0.

* XForms in XHTML validation

http://lists.w3.org/Archives/Public/www-forms-editor/2009Nov/0001.html http://lists.w3.org/Archives/Public/www-forms-editor/2009Nov/0000.html

John Boyer: Leigh, these are a good start for your action item to produce a WG Note.

Action 2009-12-2.1: Leigh Klotz to examine http://lists.w3.org/Archives/Public/www-forms-editor/2009Nov/0001.html http://lists.w3.org/Archives/Public/www-forms-editor/2009Nov/0001.html for WG Note.

Steven Pemberton: What is the URL for the Schema?
John Boyer: It's in the spec.
Leigh Klotz: It's a ZIP file.
John Boyer: The RNC one is a Zip file; the XSD one isn't. But it's not the XHTML 1.0 + XForms one you would need for the validator.

* XForms 1.1 Test Suite Issues

John Boyer: Should we defer this to December 10 or is it urgent?

** Test 5.2.1a,b,c:

http://lists.w3.org/Archives/Public/public-forms/2009Nov/0022.html http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt05/5.2/5.2.1/5.2.1.a.xhtml

Nick van: Wasn't there a problem with one type?
John Boyer: The problem was normalizedString; this form doesn't have it.
Leigh Klotz: So the theory is that you don't get xforms-valid when the processor starts but you do when you get reset?
John Boyer: According to the reset action, that sets another copy of the data into place, and you get rebuild/recalculate/revalidate/refresh. Can we find support for getting xforms-valid events in these cases? Or do we amend the test? It looks like some processors do it.
Leigh Klotz: And you said XForms said that?
John Boyer: It's a conjecture.
Charlie Wiecha: That reset behaves differently from initial form load?
John Boyer: It does behave differently but we are now more rigorous on xforms-valid. For example, you don't get the events on instance replacement; we cleared that up.
Leigh Klotz: What would a form author do? Would rebuild do it?
John Boyer: No.
Charlie Wiecha: A setvalue with empty value.
John Boyer: But it's already empty.
Leigh Klotz: So I'd never get xforms-invalid if I don't type anything?
John Boyer: In this case, empty is valid. So xforms-invalid is a fail. You should see xforms-valid.
John Boyer: reset event does rebuild/recalculate/revalidate/refresh
Nick van: http://www.w3.org/TR/xforms11/#evt-revalidate
Nick van: revalidate says "If the node changes from valid to invalid..." So if the complete document is replaced there are no changed nodes.
John Boyer: It's probably from a long time ago.
Charlie Wiecha: Under point 4, there's some suggestion of firing the events as a side effect of reset.
John Boyer: It says that the UI reflects the state of the model, not that events are the means to achieve that.
Charlie Wiecha: We need to nail down the UI discussion.
John Boyer: The link defines validity.
Charlie Wiecha: The events follow from the default action of the revalidate event. OK.
Charlie Wiecha: You have to start with non-empty process.
Leigh Klotz: Or you could use a succesful submission to test validity.
John Boyer: You'd have to check for submit-error and submit-done and look for context information for validity error.
Leigh Klotz: That sounds better than this.
John Boyer: It could start non-empty, then you could do the setvalue, then the submit.
Charlie Wiecha: It doesn't need to be non-empty.
John Boyer: It's a radically different form then. Perhaps a simpler one.
Charlie Wiecha: Yes, but the setvalue isn't necessary.

John Boyer: First, let's agree that reset shouldn't produce the events.
Charlie Wiecha: Yes, I don't like it, but that's the way it is.
John Boyer: Our upcoming UI event proposals still don't dispatch those events on reset. We have two conditions for xforms-valid or xforms-invalid: the node's value changes (UI, setvalue, bind/@calculate) or, the nodes validity changes (combination of things with revalidate).
Nick van: In the new UI event proposal, reset will have them sent when the nodes change from valid to invalid, and also for instance replacement.
John Boyer: And for form startup?
Nick van: That's not decided; there are other proposals (xforms-enabled context info, xforms-enabled).

John Boyer: OK, so what do we do for this test? Some processors are getting the event.
Leigh Klotz: The Ubiquity test shows it does get the event. The events are problematic and we know that we want to change the way the UI events work. So testing with submission is easier and less problematic.
John Boyer: It's a different test.
Leigh Klotz: The data types would be the same and that's what's being tested.
Nick van: There are several tests for validity and readonly and changing them all would be a lot of work.
Leigh Klotz: submission can't test readonly.
John Boyer: Let's look at the other tests in the chapter. We'd have to change the 5.2 tests to use submission; the others use setvalue. We're not losing coverage on these events as they're tested elsewhere. But we don't want to go through the test suite making changes.
Charlie Wiecha: We could do the same in 5.1.a.
John Boyer: That's not a contentious use of the xforms-valid event; even in the new UI system it would still work. Testing xforms-valid is good, no matter where we test it, but if we change the 5.2 tests only, then we fix the problem without materially changing how much stuff we test.
Charlie Wiecha: It's reasonable to change 5.2 to use non-empty value and setvalue.
Leigh Klotz: That works too, as long as we don't change it to something that we think is broken and want to fix next time.
John Boyer: We could change 5.1.a and use the valid values and then do setvalue to empty string.
Nick van: It should start invalid because we want invalid values.

John Boyer: Who will re-write the test?
Charlie Wiecha: I'll do it.
John Boyer: Will you also 5.2.1 a, b, and c?
Charlie Wiecha: I'll do it.
John Boyer: Submission or setvalue? If you do submission you need to look at the context info.
Charlie Wiecha: I thought I'd use setvalue. It gives finer grain results about types.
Leigh Klotz: As long as it's not a behavior we plan to change in 1.2.
John Boyer: So let's test that.
Nick van: You don't get the invalid event for string.
Leigh Klotz: The "string" test is good because it tests for the validity of the form. Perhaps the processor implementor forgot to implement xf:string type.
John Boyer: OK. So we just have to know that string won't give an invalid.
Nick van: [irc] "When you activate the Invalid Values trigger you must see an "XFORMS-INVALID" output for all the data types except string, which will either have an output of xforms-valid or no output. "

Action 2009-12-2.2: Charlie Wiecha to re-write XForms 1.1 test 5.2.1 a, b, and c to use setvalue instead of reset to respond to http://lists.w3.org/Archives/Public/public-forms/2009Nov/0022.html

** Test 11.9.8.a

http://lists.w3.org/Archives/Public/public-forms/2009Nov/0030.html http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt11/11.9/11.9.8/11.9.8.a.xhtml

John Boyer: The document encoding is iso-8859-1.
Leigh Klotz: The instance data e-with-acute-accent is iso-8859-1 and it's the same.
Nick van: It's not the same in UTF-8; it's a multi-byte character.
John Boyer: OK, so the XML file should be UTF-8 and the instance data should be in two-byte UTF-8.
Leigh Klotz: And in UTF-8 is's C3A9 LATIN SMALL LETTER E WITH ACUTE
Steven Pemberton: The W3C server is serving it up as UTF-8.
Leigh Klotz: Yes, saving it locally works.
Nick van: The DOM doesn't have an encoding.
Leigh Klotz: The header disagrees so saving and loading works.
Nick van: It's easier to fix the content than the header.
John Boyer: The test isn't broken; it's just the wrong type.
Leigh Klotz: But the LATIN SMALL LETTER E WITH ACUTE in the isntance data is in ISO-8859-1 and we must fix that as well.
Nick van: XML editors will fix it correctly.

John Boyer: Nick?
Nick van: I can do it.
Leigh Klotz: We can ask Joern to verify.

Action 2009-12-2.3: Nick van den Bleeken to fix character encoding of file http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt11/11.9/11.9.8/11.9.8.a.xhtml to be UTF-8 and adjust content � appropriately and respond to http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt11/11.9/11.9.8/11.9.8.a.xhtml

John Boyer: Maybe we will update the zip files for the tests when all of Joern's issues are done.
Charlie Wiecha: Sounds good.

** Test 7.8.2.c

http://lists.w3.org/Archives/Public/public-forms/2009Nov/0031.html

John Boyer: We don't have enough exceptions; the rules are a little snaky between the two. So which one are we supposed to get? Is there already an action to deal with this?
John Boyer: Reading the spec, I think it has to be binding exception.
Uli Lissé: [leaves]
Erik Bruchez: Value isn't a binding. The names are broken.
John Boyer: The output/@value changes if the referenced nodes change.
Erik Bruchez: There's nothing the spec that says value creates a binding.
John Boyer: "The value is updated if the referenced nodes change."
Erik Bruchez: OK, it must happen during refresh.
John Boyer: It says it in the spec. It's not rigorously defined, and a lower priority disconnect that I've been aware of.
Erik Bruchez: That's fine. Binding exception should be limited to SNB and nodeset bindings in order for people to figure it out. Then it's easy to answer the question when you get the binding exception. For MIP and binding attributes, you'd get somethign else, but we haven't madet that decisions.
John Boyer: We did; we said compute-exception is for compute and binding exception is for all others. An "at" attribte will give you a binding exception.
Erik Bruchez: It's highly unsatisfactory.
John Boyer: I agree, but we didn't get another type.
Erik Bruchez: Are we happy with the spec text then?
John Boyer: No. The lesser evil seemed not using compute if it wasn't a compute.
Erik Bruchez: It would be simpler to have xforms-xpath-error for non-binding, non-compute errors.
Erik Bruchez: Also XPath 2.0 has static and dynamic errors. We should redesign the events.
John Boyer: Nowhere can we find the comment that all XPath expression except for MIP will get an xforms-binding-exception if something is wrong with the XPath. @at, @value, a few other places that aren't SNB or @nodeset.
John Boyer: So the first thing is to fix the label in the test to say xforms-binding-exception. Nick?
Nick van: Yes.

Action 2009-12-2.4: Nick van den Bleeken to change label in http://www.w3.org/MarkUp/Forms/Test/XForms1.1/Edition1/Chapt11/11.9/11.9.8/11.9.8.a.xhtml to say xforms-binding-exception.

Action 2009-12-2.5: John Boyer to produce proposed erratum for miscellaneous XPath expressions so that all XPath expression except for MIP will get an xforms-binding-exception if something is wrong with the XPath. @at, @value, a few other places that aren't SNB or @nodeset.

Action 2009-12-2.6: Erik Bruchez to propose revampled XPath exceptions for XForms 1.2 as part of XPath 2.0 work.

* Next Meeting

John Boyer: See you next week
Leigh Klotz: [IRC] Next meeting December 9th Our last teleconference for 2009 will be December 16th, and the first will be January 6, 2010.

* IRC Minutes

http://www.w3.org/2009/12/02-forms-minutes.html

* Meeting Ends