W3C Forms teleconference May 6, 2009

* Present

John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Charlie Wiecha, IBM
Erik Bruchez, Orbeon
Uli Lissé, DreamLabs

* Agenda


* Previous minutes

* Implementation Report


Nick van: I haven't updated based on the Ubiquity changes; what has changed?
John Boyer: Ubiquity 4.2.4.a xforms-model-destruct. Leigh has tested with the Firefox plugin.
Nick van: We have to merge that in.
John Boyer: And decide to first.
John Boyer: Also we pass 10.4.f in Ubiquity+IE and Ubiquity+FireFox, for our two implementations. The issue is the interplate between the delete action and repeat index management, with the increased specificity of the specification. We have different conformance levels for IE and Firefox.
Leigh Klotz: I see IE8 is popping up, and Google Chrome is doing silent updates, so have you started testing IE8 yet?
John Boyer: Right now we're testign Firefox 3 and IE7.

John Boyer: We have 11.1.v in Ubiquity+Firefox but not checked in yet.
John Boyer: We're working on 11.9.4.b as well. The remaining ones are 8.1.10.a and 11.9.4.b, which we are working on.

* Proposed test revision for xforms-model-destruct

http://lists.w3.org/Archives/Public/public-forms/2009May/0002.html http://lists.w3.org/Archives/Public/public-forms/2009Apr/0021.html

John Boyer: Any objections to adopting Leigh's new test? None?

Resolution 2009-05-6.1: We adopt Leigh's test of xforms-model-destruct with load/@show=new in http://lists.w3.org/Archives/Public/public-forms/2009Apr/0021.html

John Boyer: It works in Ubiquity+IE; in Ubiquity+Firefox we allow the popup. The next report will show it, but we should have all of these done by the end of the week.
Nick van: Can you ask them to update the test cases for the version exception?
John Boyer: I noted that and will get it done on the Ubiquity copy.
Nick van: I'll update it.
John Boyer: Do you need an action?
John Boyer: It doesn't matter.

* Proposal to remove test of mailto with multipart-post


John Boyer: I sent mail about it and Eric responded; the issue is that xforms-submission has to work with http and https, and XML and url-encoded serialization. A reasonable way to send mail is the "mailto:" scheme in the URL of a submission, and we have two tests: 11.9.r and 11.9.p. 11.9.r uses url-encoded, so all of the instance data is attached as tag-value pairs. There are no other implementation reports other than Ubiquity on mailto. Ubiquity can't get to multipart-post. We're not testing all variations here; we're proving it's reasonable as an optional feature that you could send mail with optional submission. We two such passing tests. You just can't send mail with every type of xforms submission right now, and claim we should get rid of the test because it's an unnecessary blocker.

Leigh Klotz: If I could figure out where Chiba was right now I'd go implement it but I can't at the moment.
Nick van: I could do it too but don't have time.
Leigh Klotz: It's pretty obvious which implementations could easily provide mailto with multipart/related but don't happen to and which can't. Since it's an optional feature, I don't see the need to test every combination. mailto is clearly implementable in at least some processors.
John Boyer: We are showing an implementation of the optional feature, and we have it urlencoded-post in Ubiquity+IE and Ubiquity+Firefox. We have Ubiquity+IE working with form-data-post as well.

Resolution 2009-05-6.2: We remove test for optional feature optional feature 11.9.p because the optional feature itself is tested in 11.9.r, and rename 11.9.r to 11.9.p.

Action 2009-05-6.1: Nick van den Bleeken to remove test for optional feature optional feature 11.9.p because the optional feature itself is tested in 11.9.r, and rename 11.9.r to 11.9.p by next week.

* Updating FF Plugin report for 8.1.6.b


John Boyer: Nick, can you do this now that Keith agrees with Leigh that Firefox passes thsi test?
Nick van: xforms-value-changed on upload?
Nick van: I think I already did it, but if not I'll do it.

* Action Item List


John Boyer: Are there any test-suite action items we can defer?

Leigh Klotz: My Action 2009 03-18.1 is done.

Charlie Wiecha: I have some for Appendex B (9, 11, and 14 tests).
John Boyer: I have those too.
Charlie Wiecha: I took them but you suggested it instead, so I'm listed incorrectly.
Charlie Wiecha: There's also the xsi:nil test.

* Other Test Suite Issues

Nick van: I need to update the license.
John Boyer: Not until we're ready to publish.
Nick van: I've updated the colors.

John Boyer: Converting the drivers to XHTML (Uli) is nice but not necessary for this week.

John Boyer: Leigh, any test-suite related actions?
Leigh Klotz: None I could see.

John Boyer: You have a test b.14 action as well, probably a duplicate.
Erik Bruchez: I have a bunch of them, including that one.
John Boyer: The b.14 one will probably get done by someone else.
Erik Bruchez: We need to fix the ref and and add mediatype.
John Boyer: ...
Erik Bruchez: ... group label ...
John Boyer: OK, tell the group which changes you make since Ubiquity has a copy.
Leigh Klotz: You could use XSLT.
John Boyer: We have to hand-edit the results of the XSLT.

* Target id

John Boyer: For the targetid issue, how do we want to handle that from the test suite perspective?
Nick van: We decided that I was going to add the attributes to the tests and make the other tests not required.
John Boyer: How many tests? Do we have to update all of them? Do we just make sure that targetid is represented?
Nick van: Tests that depend on optional or deprecated attributes.
John Boyer: I'm having misgivings about this change, so late in the process. There may be issues with Ubiquity+IE or Ubiquity+Firefox.
Nick van: ...
John Boyer: But I guess we'll have to go forward with those.
Leigh Klotz: Is that any change from status quo?
Nick van: We recapped what we said last week.

* 11.2 replace=all

John Boyer: I have action item about replace=all errors. Did we change the tests to say you don't get them on a replace all?
Leigh Klotz: I have a vague recollection that we dropped a test, perhaps at the F2F.
John Boyer: I see; some tests have already been changed.
John Boyer: I see some tests I need to confirm are the latest ones.

* Face to Face Agenda

John Boyer: It would be helpful if we worked on 1.2 and 2.0 themes and requirements rather than attacking a laundry list. We can pass this to the rechartering, as our charter expires at the end of the year.
Charlie Wiecha: Move vote would be to make the transition away from Forms and into the RIA space. We could also discuss extensibility to HTML6. It might be nice to see if we can get Sam Ruby to join for conversation by phone.
John Boyer: Interestingly, we also have our Virtual Day June 4th.

John Boyer: Steven asked that we consider the rechartering. There's XForms for HTML as well.
John Boyer: I'll ask for a sign-up sheet.

Charlie Wiecha: Mark was talking about a Tuesday evening event.
John Boyer: Maybe we can end early that day.

John Boyer: If there isn't going to be an evening event, then holding it in Amsterdam would make more sense as we get Steven.
Nick van: It's getting a bit late.
Charlie Wiecha: He sent a note a week or so ago.
Nick van: For me, it was vague.
Charlie Wiecha: Yes, we need to follow up on that.

John Boyer: For 1.2/2.0, merging recalculation and revalidation seems huge, but putting src on model doesn't.
Charlie Wiecha: Structural transformation in binds, for Backplane.
John Boyer: It's on the 2.0 list.
Charlie Wiecha: I'm revoting. And the notion of applying the data-dependency graph document-wide.
Leigh Klotz: To the UI?
Charlie Wiecha: Yes.
John Boyer: Those sound like the gray area of the non-fine-line of 1.2/2.0.
Charlie Wiecha: They feel less far out because others people are doing them. Structural transformation might affect the rebuild loop. But document-wide constraint functions is getting more real all the time.
John Boyer: It's re-using the same concepts. The atomic operations are still going to be expressed in stuff we already have. If you did the equivalent of a calculate to attach a tree, it would probably still involve an insert.
Charlie Wiecha: Or if you use xquery update, how do you track the dependency logic?

John Boyer: So let's talk next week and decide how rigorously we want to define 1.2 and 2.0 in the charter document.

Leigh Klotz: We spend a lot of time saying we're doing small things and they turn into medium things. We could do like HTML5 and say we're doing one big thing for three years, and then the publish other documents in the interim.
John Boyer: Let's talk next week.

* IRC Log


* Meeting Ends