W3C Forms teleconference May 2, 2007

* Present

Blake Jones, ViewPlus Technologies
Charlie Wiecha, IBM
David Landwehr, PicoForms
Erik Bruchez, Orbeon
John Boyer, IBM (co-chair)
Joern Turner, DreamLabs
Leigh Klotz, Xerox (minutes)
Mark Birbeck, x-port.net
Nick van den Bleeken, Inventive Designers
Steven Pemberton, W3C/CWI (co-chair)
Dave Raggett, W3C/Volantis

* Agenda


* Announcements

John Boyer: My voice is going; Steven can you chair?
Steven Pemberton: OK. I gave a talk on XForms at Apachecon today, and the evolution of programming languages.

* Coming calls

May 16 Sebastian to chair telecon May 23 Steven to chair telecon

Steven Pemberton: Call next week? It's the web conference. John will you be at the web conference?
John Boyer: Yes.
Steven Pemberton: So we have a call next week. I understand it is -20C in Banff.

* Next FtF - Hosted by IBM June 13 to 15, 2007

Charlie Wiecha: Can you post the link again? Our internal page still points to Palo Alto.
Leigh Klotz: It also still points to the old charter.

Steven Pemberton: http://www.w3.org/MarkUp/Forms/Group/ is the old page. It's now http://www.w3.org/MarkUp/Forms/Group/
Charlie Wiecha: That's the public page.
Steven Pemberton: We don't have an internal page anymore.
Nick van: [irc] http://www.w3.org/MarkUp/Forms/2007/06/ftf/

Steven Pemberton: Please fill in the form, ASAP.

* XForms 1.0 Third Edition

Steven Pemberton: Anything to say?
John Boyer: I'm working on it. I have starred the things we need to talk about.

* XForms Basic 1.0, 1.1

Leigh Klotz: I need to do the implementation report. I will mail the PR to the folks at PicoForms.
Steven Pemberton: If you can get this done before next week I have a meeting with Steve Bratt to schedule a transition call.
Leigh Klotz: I will take the XForms 1.0 one, remove the Schema stuff, change the name.
Steven Pemberton: And add the photo?
Leigh Klotz: Steve Bratt didn't like the photo before.
Steven Pemberton: OK, so no photo.
Leigh Klotz: Just letting you know.

* XForms Conference at XML Conference

John Boyer: I spoke to Dave Megginson and he was hopeful.

* Forms Joint Task Force with HTML WG

John Boyer: I think Dave Raggett won't be involved because he's chairing a new group. We need some people to participate in this task force. This is for XForms Transitional and provide feedback on whatever that turns into. Should we poll for interest?
Steven Pemberton: Sure.
John Boyer: Dan Connolly has done something similar for the HTML WG and 6 or 7 people have expressed interest.
Steven Pemberton: I'll look at his questionnaire and make one.
John Boyer: His is part of a bigger one; I'll send you sample text.

Action 2007-05-2.1: John Boyer to send sample text for HTML WG Joint Task Force interest to Steven Pemberton.

Action 2007-05-2.2: Steven Pemberton to create questionnaire for interest in HTML WG Joint Task Force.

* Issues with setvalue and context of @value


Steven Pemberton: It's not a last call issue because it was sent to www-forms.
John Boyer: I'm amenable to discussing it now or later. I think this did end up becoming a last-call comment; I'm not sure. It seemed a bit of an inconvenience but here's something you can't do.
Leigh Klotz: So current doesn't work here?
Mark Birbeck: No.
John Boyer: That's the output of the ref,not the context of the ref.
Mark Birbeck: There's loads of use cases where you have to use the index of the repeat.
John Boyer: Is that a workaround?
Mark Birbeck: Yes. The trigger changes the current index of the repeat, if you believe the action handler is run after the index is set.
Leigh Klotz: The same happens with labels inside of trigger with ref all inside of repeat. You can't get it back with that hack.
Erik Bruchez: [irc] We have implemented such a context function
Erik Bruchez: [irc] Variables would also solve this problem
Nick van: [irc] an action with a while on it, but then it is a repeat again.
Erik Bruchez: [irc] context() should could rather take the id of the enclosing element
Mark Birbeck: Nor with this proposal. All you've done is defer the problem. Unless you have a context function, there
Leigh Klotz: Some can be done by static analysis and repeating the expression, but repeat can't. You may not need the arbitrary context.
Mark Birbeck: And action handlers.

Steven Pemberton: So what do we do?
John Boyer: We give the index solution for now, and say that we have added the general case to our futures list (e.g. Leigh's issue).
Steven Pemberton: OK. Who will?
Charlie Wiecha: I will.
Mark Birbeck: I see using index came up in the thread already.
Charlie Wiecha: Yes.
Mark Birbeck: There should be a neater solution.

Action 2007-05-2.3: Charlie Wiecha to reply to http://lists.w3.org/Archives/Public/www-forms/2007Apr/0027.html with the index solution for now, and say that we have added the general case to our futures list.

* encoding, charset and SOAP


Steven Pemberton: [reads]
John Boyer: We need someone versed in submission issues to take this offline.
Mark Birbeck: I know we've implemented this; I'll ask Paul, who's done this, to read it.
Charlie Wiecha: It may be a SOAP issue rather than an XForms issue.
Mark Birbeck: I guess it's that we've got two ways to

Action 2007-05-2.4: Mark Birbeck to check his implementation decisions on http://lists.w3.org/Archives/Public/www-forms/2007Mar/0060.html

* Submission pruning empty nodes on url-encoded submissions

Leigh Klotz: This appears to have the wrong link in the agenda.
John Boyer: I'll find the right one later.

* Readonly trigger is disabled?


John Boyer: I wasn't aware that a trigger that was readonly would somehow behave as if it were disabled. The description says "grayed out."
Leigh Klotz: We've had use cases for displaying greyed-out unselectable select items with hints that display and the hint says why.
Erik Bruchez: We have two use cases; a trigger hidden or shown, and another shown but greyed out. The problem is that if both match relevance you need to resort to style. That's less convenient than having two different MIPs. That's the initial rationale for saying a readonly trigger is there but greyed out; consider a readonly input field; you are fairly likely to display the contents of the input field, but you are likely to grey out the field and make it not focusable, whereas if it is non-relevant it would be non-visible or not show the data to which it is bound.
John Boyer: I can relate more to the way you might choose to display by default. But for textarea, you might make it readonly to show what someone else has typed in in a previous stage of a workflow. For example I might get readonly fields where someone else has filed the vacation request and I just read and approve.
Mark Birbeck: Greyed out usually means not relevant in operating system dialog boxes; it's an indicator of what you might get if you tick the box above. They're read-only yet in our model you're going from non-relevant to relevant. You have to be careful.
Erik Bruchez: How do people usually implement relevance? I know you can style them any way you want in theory, but what do people do? By default we don't display it at all. This applies to groups as well, to emulate XForms switch, the ref=".[]" trick for example.
Mark Birbeck: So you asked what do people normally use?
Erik Bruchez: So a trigger that is non-relevant is not displayed at all.
John Boyer: In our case, if you turn the visible property on, they show.
Erik Bruchez: That's not in XForms.
John Boyer: Yes, that's in our host language.
Mark Birbeck: By default...they're hidden. That's not the only way. We have forms where we use a different technique.
Leigh Klotz: What's issue here? It's not about the presentation because CSS or as John points out the host language is in charge.
Mark Birbeck: So people think one thing about input, and at least two things about textarea. But what does readonly trigger mean?
John Boyer: If you set a readonly MIP on a node the trigger is bound to, does the trigger not work and does it look disabled?
Mark Birbeck: It's hard to suppress that the control can suppress events. Where do you stop? Can the output suppress keypresses? When a span has focus, the keypresses get reflected through the DOM. So why not output? As for the appearance, that's different; I wouldn't have a big objection if we added to our suggestions for consistency, as there's no requirement.
John Boyer: If it appears to be disabled it would be problematic if it worked.
Mark Birbeck: To quote David Landwehr, it doesn't create non-interoperable forms. A voice system would do it differently anyway.
Erik Bruchez: It would create non-interoperable forms.
Mark Birbeck: We don't have the notion of disabled in XForms. We don't say that readonly input should not propagate the onclick event.
John Boyer: For a readonly node, the only thing we concretely say is that the input won't modify the node. We kind of assume that that roughly means that the user won't be able to enter any character data into that input.
Mark Birbeck: But you could easily come up with a use case with a Google map and a pin in the middle and allow the user to drag and zoom; the control isn't disabled, but you can't modify the data.
John Boyer: Or page up and down in a textarea. It is readonly in the visual UI sense.
Mark Birbeck: I have a problem with trying to say too much. You're right; we say can't modify the model but beyond that we can't predict all the UI use cases.

Steven Pemberton: Do we feel that we resolve the issue? Erik?
Erik Bruchez: I was answering John's email. I have an opinion contrary to John's and Mark's. I'm sensitive to the separation argument, so I'm fairly inclined to agree with that. I remain fairly ...
John Boyer: What if it has an action that does just a send or a setfocus?
Erik Bruchez: It doesn't have a value. If an implementation allows scrolling in a readonly text area, fine. The input stores data and shows it. But for trigger, the only reason you have it is to click and the only reason to bind data is to leverage MIPs.
John Boyer: I'm halfway between Erik and Mark; I thought you couldn't activate the trigger, but not on the display.
Mark Birbeck: I wouldn't disagree if we were talking about relevance. But readonly, I think its a stretch. It does seem odd that readonly would have an effect; do we then move onto invalid? We fixed it lately so that non-relevant controls aren't there, so they are disabled.
Erik Bruchez: Yes, non-relevant controls are effectively disabled. So it's a little funny that they can be styled.
Mark Birbeck: It's not a real control; it's a picture, a snapshot.
John Boyer: It's not that it's not there; it disconnects its event handlers.
Mark Birbeck: That's much more what disabled means to me.

* IRC Minutes


* Meeting Ends