W3C Forms teleconference Setember 8, 2010

* Present

John Boyer, IBM
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Philip Fennell, MarkLogic
Steven Pemberton, CWI/W3C (chair)
Charlie Wiecha, IBM
Alain Couthures, AgenceXML
Uli Lissé, DreamLabs

* Agenda


* Previous Minutes

* US Government FCC REST APIs


Leigh Klotz: They're using WADL, XML, REST, JSON, etc. This is part of the data.gov initiative. It might be useful for doing some examples.
Steven Pemberton: I switched over to caldev for calendars. We don't yet support that format, like vCard. Perhaps we should consider that as well.
Leigh Klotz: I wonder if this might be Kurt Cagle as well.

* Lyon


Steven Pemberton: Now is the time to register. There will be 2 or 3 there and the rest on the phone, Monday, Tuesday, Thursday, Friday. TPAC day is Wednesday.

* Action Item Review


John Boyer: I reported that I've started the wiki page for 1.1 errata. I've written up two of the errata, using the diff-marked wiki language. http://www.w3.org/MarkUp/Forms/wiki/XForms_1.1_First_Edition_Errata

* XForms Wiki Spec

Steven Pemberton: I have put XForms 1.1 into the wiki. It's not perfect. The internal links aren't working. It looks quite good.
John Boyer: And the other direction?
Steven Pemberton: Yes, I have the Python files, written by someone else. They're free with no support. I'm working on it now.
Charlie Wiecha: HTML or SpecXML?
Steven Pemberton: The code produces HTML but it seems not hard to produce SpecXML once we get that working.
John Boyer: There are a couple of blank examples, but it looks fantastic.
Nick van: How about the diff marking.
Steven Pemberton: That's the next step. We could produce diffs from specxml. Shane has a program that does that.
John Boyer: SpecXML has diff.
Steven Pemberton: I'm deeply impressed that you've done that by hand, but it's a pain and error-ridden, so automating it would be a good process.
Charlie Wiecha: Doesn't the wiki have a layering delta mechanism.
Steven Pemberton: You can see the changes. That might be enough, to collect two versions and compare them. It's not beautiful though.
Charlie Wiecha: That could be the input to Shane's comparison.
Steven Pemberton: I think Shane's comparison takes two HTML things and does a smart diff, though I don't know how.
Leigh Klotz: So we don't lose any data by not diff-marking, as it's non-normalized data, an the wiki diffs give us the interim results.
John Boyer: There are a couple of blank examples, section 8.1.1, just before 8.1.2. Also 5.2.1.
Steven Pemberton: I will look at it. I was surprised at how relatively easy it was.
John Boyer: It sounds like a browser issue, so it's not a problem for this version as it's not normative.
Steven Pemberton: Yes, it's the editor's version.
Charlie Wiecha: We need to get people to use that pattern in the wiki.
Steven Pemberton: Yes, it has to be documented.
John Boyer: In IE7, there aren't any black boxes.

Steven Pemberton: The variables drive the next-generation of document expressions in Python. I'd like to make that XML to use XForms to configure the variables.

* Update on NVDL for XHTML+XForms


Philip Fennell: There aren't any action items for me on this. I've distributed what I've done. I've run this against the XForms documents in the test suite and found some issues. One set is class, navindex, tabindex. I'm trying to find a way to get the nvdl validation to work, but haven't finished it. There are some issues with issues that come up invalid against the schema. On the whole it seems quite effective for authoring XHTML+XForms.
John Boyer: Is there a list of what's broken in the tests?
Philip Fennell: Yes, in the email and then a README file in the zip with the log output. Some are errors because of testing errors. Others look like mistakes, two or three.
Steven Pemberton: Can we discuss class/accesskey/navindex?
Philip Fennell: Those have been taken out of the XML schema and said they're host language. Is there a way in NVDL to re-attach those? When they appear without namespace prefixes, the assumption is that they're bound to XForms and throws an error. There are ways to define specific contexts for matching, which might get around the problem.
Steven Pemberton: I suspect that we'll have to add these back into the schema.
Leigh Klotz: If it's an XHTML integration issue we should suggest adding it to the XHTML in integration, not the basic schema.
Leigh Klotz: Philip, are you using RNG or XSD schema?
Philip Fennell: XSD
Leigh Klotz: Owen Newnan has a bunch of fixes sent to us for the XSD schema. I updated the RNC schema and it validates the test suite.
Philip Fennell: I should take a look.
Steven Pemberton: What are the next steps?
Philip Fennell: From my point of view, I'll try the RelaxNG schema and see how that behaves. I'd like to somehow re-attach the attributes to the HTML schema using NVDL.
Steven Pemberton: It's something we've long talked about in XHTML Modularization. But I don't think the XSD schema uses that. That's one way of inserting attributes into a Schema for different host languages.
Nick van: I think we should also go through the error messages.
Leigh Klotz: I think we should fix the schema before. Here's the link to Owen Newnan's report on our XSD: http://lists.w3.org/Archives/Public/public-forms/2010Mar/0001.html
Philip Fennell: I can certainly do that.

Action 2010-Setember-8.1: Philip Fennell to read Owen Newnan's report on our XSD and propose changes: http://lists.w3.org/Archives/Public/public-forms/2010Mar/0001.html

Leigh Klotz: We already know the RNG schema validates the test suite and we know the XSD doesn't validate it.
Steven Pemberton: So we can update it outside TR space.

Action 2010-Setember-8.2: Philip Fennell to integrate RelaxNG schema into NVDL validation.

* How to respond to xforms-out-of-range


Leigh Klotz: out-of-range and invalid aren't related to each other. out-of-range is presentational.
John Boyer: I think empty string is different from a non-listed value.
Leigh Klotz: I don't think it's different. Empty string is out of range for a checkbox and it's out of range for a select1 which doesn't include empty string.
John Boyer: I think it's different because empty string is an initial value.
Nick van: ...
John Boyer: If there is no match and and the form control is not empty it should displaya as out of range.
Leigh Klotz: I didn't know that we had special text about empty string in select1.
John Boyer: http://www.w3.org/TR/xforms11/#ui-selectOne

For closed selections: If the instance data matches the storage data of one of the selection items, that item is selected. If there is no match, no items are initially selected. If there is no match and the storage data is non-empty, the form control must indicate an out-of-range condition. If the form control switches to or from being out-of-range, then xforms-out-of-range or xforms-in-range must be dispatched to the form control.

John Boyer: For select1, empty string is not out of range.
Leigh Klotz: We should say that out of range is presentational and invalid is model.
John Boyer: That's the right answer; a constraint says it's invalid but select1 can show it.
Leigh Klotz: I think we've discussed it.
Steven Pemberton: John, would you answer?

Action 2010-Setember-8.3: John Boyer to answer http://lists.w3.org/Archives/Public/www-forms/2010Sep/0007.html about out-of-range and invalid.

* XForms binding exception (clarification?)

http://lists.w3.org/Archives/Public/www-forms/2010Jun/0005.html http://lists.w3.org/Archives/Public/www-forms/2010Jun/0006.html

Steven Pemberton: We use the word "illegal" just once in 4.5.1.
Leigh Klotz: I've seen submission/@ref="false()".
Steven Pemberton: We clearly have a task to define it. Do any implementors here send this event?
John Boyer: Yes.
Steven Pemberton: For illegal binding expressions?
John Boyer: Yes.
Steven Pemberton: What does that mean?
John Boyer: Syntactically invalid XPath gives binding or compute exception. In a nodeset or ref attribute and you get something that isn't that type, we throw and exception for that too. I think we talked about this.
Charlie Wiecha: I think so.
Steven Pemberton: I'll send a note to the group and we can kick off the discussion of when this is sent.

* IRC Minutes


* Meeting Ends