W3C Forms teleconference November 28, 2007

* Present

Charlie Wiecha, IBM
Steven Pemberton, CWI/W3C
Uli Lissé, Dreamlabs
John Boyer, IBM (chair)
Keith Wells, IBM
Sebastian Schnitzenbaumer, Dreamlabs
Erik Bruchez, Orbeon
Roger Pérez, SATEC
Leigh Klotz, Xerox (minutes)
Nick van den Bleeken, Inventive Designers
Blake Jones, DAISY/ViewPlus
Mark Birbeck, x-port.net

* Agenda


* Previous minutes

* XForms 1.1

John Boyer: The announcement is supposed to go out tomorrow. Have any AC reps received the decision yet?
Steven Pemberton: The decision is released on the day of publication.

* XForms Conference at XML Conference


Charlie Wiecha: Any thing to discuss?
Leigh Klotz: I saw Noah Megginson announced the XForms session on the xml-dev.

* Action item list triage


Nick van: I need some confirmation of the action item list triage.
John Boyer: Also everyone please clean up your action list.
Steven Pemberton: After vacation.
John Boyer: And is Mark Seaborne on?
Steven Pemberton: On IRC.

* Action on public-forms causing perceived lack of action on www-forms

John Boyer: I got a note from Aaron Reed indicating that the www-forms list looks like it's not active. But the perception is that the activity of the group has gone down.
Leigh Klotz: Should we be posting a monthly summary of what's been happening? I'd be happy to that.
John Boyer: How about cross-posting minutes?
Leigh Klotz: We can summarize what has happened in the past month; if people want actual minutes, they can read the public-forms list. Providing a higher-level view may have an advantage.
Steven Pemberton: It's weird having two public lists, but people aren't allowed to subscribe to the working-group list.
John Boyer: That's a big part of the problem. We can't move them over, unfortunately.
Charlie Wiecha: So you can't subscribe to the public-forms list?
John Boyer: No.
Steven Pemberton: You can't post but you can read via RSS.
Erik Bruchez: [irc] then we should send instructions to www-forms about how to subscribe to the RSS feed
Nick van: [irc] for all lists there is a link to the RSS feed on the homepage but NOT for public-forms@w3.org
John Boyer: So we should update the home page.

Action 2007-11-28.1: John Boyer to update home page with link to RSS feed for public-forms.

Action 2007-11-28.2: Leigh Klotz to summarize group activity monthly to www-forms, starting this month, and include pointers to www-forms RSS feed and archives.

* Question about while attribute

http://lists.w3.org/Archives/Public/www-forms-editor/2007Nov/0004.html http://lists.w3.org/Archives/Public/www-forms/2007Nov/0000.html

John Boyer: The observation is that it doesn't seem to be useful for most actions, because you really only do them once.
Nick van: [irc] uli replied, to it. when you have async submission, a bit of a stretch
John Boyer: http://lists.w3.org/Archives/Public/www-forms-editor/2007Nov/0004.html
John Boyer: I don't know that that's the reason.
Steven Pemberton: For any action you can wrap it with an xf:action and apply it there, so there's no sense in disallowing it. Also, what makes sense can change over time. We noticed this with HTML4; we talked about which elements should get the id attribute, but then we discovered we needed it everywhere.
Erik Bruchez: With setfocus, couldn't setfocus cause an event to be sent while being processed? There could be side effects.
John Boyer: Yes, that's what I thought was a good technical answer.
Erik Bruchez: Yes, so we'd have to be really certain that an action had no side effects. So I think Steven's approach is better.
Nick van: And delay.
John Boyer: I had thought we'd still have implementations not interrupt currently-running action sequence in order to complete their jobs.
Uli Lissé: I understand; I hadn't thought about it.
Mark Birbeck: I agree with other comments; if we want to go in a more modular direction in the future, these might be a separate spec anyway, SO you might have a @while on a SMIL action.
John Boyer: Many actions you think don't have side-effects do because of event handlers. As Steven says, It's academic anyway because of the action wrapper, and as Mark says, the modularity issue. Uli would you summarize it?

Action 2007-11-28.3: Uli Lissé to respond to http://lists.w3.org/Archives/Public/www-forms/2007Nov/0000.html

* Offlist ping about answer for charset and encoding (action MarkB on May 2)


Mark Birbeck: I'll have to check this out.
John Boyer: This week?
Mark Birbeck: I'll ask now.
Steven Pemberton: I don't know much about this, but why isn't it in the XML declaration at the front of the SOAP envelope?
Mark Birbeck: How do you put that in the instance?
Steven Pemberton: A SOAP package is an XML document; the XML document says what the encoding is.
John Boyer: The XML declaration in the instance isn't preserved; there are several submission attributes modeled after XSLT to control the serialization. He's saying that submission/@encoding sets the character set in the XML declaration.
Steven Pemberton: Oh, I misunderstood.
Mark Birbeck: I thought it was a reference to the HTTP headers; you might have two different charsets in two different headers.
John Boyer: The content-type header.
Mark Birbeck: The encoding sets the charset and the mediatype sets the content-type. What if you have both? I think this is a SOAP question; I would think the mediatype would win out, but I'm asking.
John Boyer: The spec says the encoding attribute affects serialization. The mediatype attribute says it doesn't affect serialization. The issue is that I wasn't aware that encoding affected the content-type. Now that we have a header element you could add a content-type using the element. In the past we said we'd just let the user agent work it out. It may be good to have some other implementers look at this. These are part of the group of four taken from XSLT output.
Mark Birbeck: The reference to "some processors" is for Formsplayer, and we do indeed append; however, mediatype wins out, because the spec is explicit about mediatype and isn't explicit about encoding. So there's only one by the time it gets to the server, but that's because it isn't explicit on encoding.
John Boyer: So we need a separate discussion about whether encoding should do that. If you have application/xml as mediatype, the XML itself is self-determining.
Mark Birbeck: No, we need to clarify it; you need the harmonization. If you sent it as a MIME package, it can be whatever you like, but if you use the XML MIME types I think you have to be more specific.
John Boyer: We need to do a little spec tightening.
Mark Birbeck: I'll read this more after the call.

* Administrative Issue

Sebastian Schnitzenbaumer: We have a visitor on the IRC channel; is this allowed?
Steven Pemberton: Yes, as long as they don't speak during the meetings.

* XForms 1.1

Steven Pemberton: So XForms 1.1 has a new shortname, xforms11. So the latest version of XForms doesn't take you XForms11. I'm not sure what we should do about that?
John Boyer: Is that what it supposed to do?
Steven Pemberton: That's partly up to us. It means people who go to W3C who go to http://www.w3.org/TR/xforms and latest version, never get XForms 1.1.
John Boyer: Should the latest-version link produce things that aren't recommendations?
Steven Pemberton: That's also a good question.
John Boyer: If I go to TR/xslt then I get XSLT 1.0 from 1999. I don't know what I'd type to get XSLT 2.0.
Leigh Klotz: You type "saxon."
John Boyer: And if you want XPath 2.0 you type xpath20.
Steven Pemberton: For SVG, you wind up at SVG 1.1. So I think we should be asking for some redirection so that we have xforms, xforms10, and xforms11.
Nick van: So we still want the latest version of xforms10 pointing to...
Steven Pemberton: Let's see what TR/html.... Ah it has different latest-version links in the specification.
Leigh Klotz: So we can't go this for XForms 1.0 but we can do that for XForms 1.1.
Steven Pemberton: So the /xforms link goes to the most recent version.
Leigh Klotz: And in that version we can have the disambiguation section at the top.
Steven Pemberton: So we create a shortname xforms10 and ask for xforms to go to xforms11.
John Boyer: Can we pass pubrules with two latest-versions?
Steven Pemberton: Pubrules check is a machine anyway; you just say the reason for not passing.
John Boyer: I'll have to rewrite the XSLT.
Leigh Klotz: So in XForms 1.1 it's a link to the TR/xforms10, not a link to a specific old version.
John Boyer: Right. I'll add this to teh issues list.
Steven Pemberton: I'm doing it now.

* Triage of future features


John Boyer: We need to decide XForms 1.0 vs. XForms 2.0.
Nick van: [irc] I made a section XForms Simplication
John Boyer: Maybe we should start with those.

John Boyer: Make the model optional.
Mark Birbeck: The model element. It's about making the page simpler.
John Boyer: Could you have multiple instances?
Mark Birbeck: That's the kind of thing we need to discuss, but I was thinking of two levels of development. A new developer wants to do some basic stuff but with a little extra, maybe input first name, input surname, and then an output. I also suggested the form element giving you one model, one instance, one submission. Not taking the instance out of the model. Then gradually introduce the MVC model. If you're loading an instance from another location, then at that point it's the model element.
John Boyer: How do they make the implied instance explicit?
Mark Birbeck: ...
John Boyer: How about just the instance element?
Mark Birbeck: What you're talking about is something like a data island. So we could break that out, instance inline or with source. Then we can break that out into a modular spec. It could be serialized using JSON as well. I'm not sure where to draw the line. I'd like to turn XForms into 20 minutes, but I don't know where to draw the line. So for example, you could define the instance element, and a couple of APIs. That's a useful module for many situations.
John Boyer: It's hard to avoid some modularization, where we end up with instance, submission, and bind as modules. Instance and submission seem to have use in implied model with a form tag. The goal of the simplification is to provide that onramp.
Mark Birbeck: I would put in message. A lot of them let you instantiate messages in different ways. You can give it a DOM entry. That's the message element. We've got lots to offer there. These low-level features don't require lots of XForms. You wouldn't even need the xpath part.
John Boyer: The final thought might be with multiple instances and one of them implied, how do we move data between the declared instances and the implied.
Steven Pemberton: If there is an implied instance, there can be no other implied ones.
Mark Birbeck: But what about declared ones?
Steven Pemberton: The moment that there are declared instances you lose the implied one.
Leigh Klotz: That makes it hard to keep something working while you move.
Steven Pemberton: So people will learn.
John Boyer: What actually breaks?
Steven Pemberton: The ability to refer to it.
Mark Birbeck: The first instance rule could go.
Steven Pemberton: That's a big change.
Mark Birbeck: We could change it to refer to the implied instance; otherwise the first instance rule isn't that useful once you cut and paste anyway.
John Boyer: Another way to put it is that when you have an implied model, the implied instance is the first one.
Leigh Klotz: If you have an unidentified first instance then you don't have the implied instance?
John Boyer: Good. Anywhere.
Mark Birbeck: So it's unnamed, not implied.
Steven Pemberton: I don't like it yet. It changes the meaning of existing forms.
Mark Birbeck: The instance() function with no args looks like you're referring to the unnamed instance.
Steven Pemberton: In XForms 1.0 ref="foo" is the first instance and in 1.2 it would be the invisible instance.
Nick van: [irc] if you have xforms 1.1 then those will not work in 1.2
John Boyer: Good point Steven.

* Meeting Ends

* IRC Log