W3C Forms teleconference January 9, 2008

* Present

Charlie Wiecha, IBM
Erik Bruchez, Orbeon
John Boyer, IBM (chair)
Leigh Klotz, Xerox (minutes)
Mark Birbeck, x-port.net
Nick van den Bleeken, Inventive Designers
Steven Pemberton, CWI/W3C
Uli Lissé, DreamLabs
Keith Wells, IBM

* Agenda


* Previous Minutes

* Next F2F


John Boyer: The virtual day is Feb 1 and the three-day meeting is Feb 4-6.
Keith Wells: The details page is on my priority to-do list.
Steven Pemberton: The virtual day is the Friday and the real meeting is the Mon-Tue-Wed.
Charlie Wiecha: Can we decide the time for the virtual day? I'll be traveling.
John Boyer: So do we have them in the meeting time zone, or the east coast?
Steven Pemberton: I'll be in the US on Friday anyway.
Charlie Wiecha: Yes.
John Boyer: I propose 9-5 Eastern time zone. That seems hard for Europeans.
Nick van: [irc] I'(l be traveling to the states on Saturday
Steven Pemberton: For Europe that's 2, 3, or 4 in the afternoon. So it wouldn't encroach too much on bedtime.
Nick van: It is the day before I start traveling; if it stops at 11 at the latest...
Steven Pemberton: We could have it one hour earlier; that's awkward for the west coast.
John Boyer: 5am is too early. We could consider doing six hours.
Charlie Wiecha: That would help those traveling on Saturday.
John Boyer: Let's do that then.
Charlie Wiecha: Can we use Skype or are we using Zakim?
John Boyer: I thought we were using Zakim. Steven?
Steven Pemberton: We can use Zakim. A virtual whiteboard as well as the IRC channel would be good.
Charlie Wiecha: We can try a web conference; we have one outside the fire wall. I can call Zakim.
John Boyer: Anything else?
Steven Pemberton: I should have an action to produce the F2F page, or we can do it on the wiki.
Uli Lissé: [irc] What about hotel reservation?
John Boyer: That's what we want Keith to look into.
Keith Wells: [irc] Yep -- will do
John Boyer: Do we need the questionnaire?
Steven Pemberton: We do. Maybe Keith can produce the page on the wiki.

Action 2008-01-9.1: Steven Pemberton to prepare questionnaire for next F2F.

Action 2008-01-9.2: Keith Wells to produce F2F info page in wiki based on previous F2F page.http://www.w3.org/MarkUp/Forms/wiki/FaceToFace
Nick van: [irc] this is the info of the previous meeting in Raleigh : http://www.w3.org/MarkUp/Group/2005/09/f2f/

* Backplane


Charlie Wiecha: I updated the patent policy and sent it off to those who might be listed as direct authors; once we get feedback I'll post it to the various lists. I think Leigh and Steven in addition to IBM folks thought they would be useful. We need three members. The main changes are the patent policy for the XG for RF intent with exclusions. People have to say at the outset of the XG rather than at the end. I also suggested weekly telecons; it seemed there were enough use cases and examples for submission, data models, and events that we could discuss that with some of the other WGs. It's a one-year thing; it can be renewed. So let me know if it's OK to post by Friday.
John Boyer: Any other agenda items, please speak up at any time.

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


John Boyer: Aaron Reed was asking about this.
Mark Birbeck: We've had some problems to sort out; I'll try to work on this week.
John Boyer: Maybe two weeks?
Mark Birbeck: OK.
John Boyer: I tried to triage some of your action items.
Mark Birbeck: Thank you. I thought that was an excellent summary; thank you. My apologies.
John Boyer: Nick, please handle Mark's action items from that message.

* Definition of card-number datatype


John Boyer: This came up as a last-minute problem for CR. We still have an outstanding reference to ISO-7812, which is an error. We had resolved to get rid of the reference. But in so doing, I started trying to find some other way to rationalize the length limit; we said it had to be 12-19 digits, which is what's specified in the Schema definition. The problem with that is the only reference we make is the Luhn patent, which has no explicit length restriction on the number of digits. In fact, it turns out that one of the big consumers of the Luhn algorithm is the Canadian social insurance number which is nine digits long. So I think the solution is to reduce or eliminate the restriction on the length for cardnumber. So it's really a datatype that can be used by the UI to allow spaces or dashes between certain groups of number.
Leigh Klotz: In the first example for that I had allowed space and dash in the lexical space, but this is better.
John Boyer: If we were to reduce that restriction, you have to have at least two digits. It seems like you might have to have three if you read the patent. Is there any objection to getting rid of the maximum length requirement? OK, none. And minimum two? Is that a significant change for CR?
Steven Pemberton: I don't think so.
John Boyer: I think we should report it.
Steven Pemberton: That's for sure. It's a CR issue; it's been determined by someone using it.
John Boyer: Since we're loosening a restriction, it won't cause existing documents to fail. I propose length restriction of min=2, no max. Any objections?
Steven Pemberton: If people do use that for credit card numbers, it won't catch credit cards any more.
John Boyer: It's already not that good. It's between 12 and 19 and my credit card is 16.
Steven Pemberton: So you need a restriction anyway. Maybe we should point it out in the description of the datatype.
John Boyer: It's already card number, so it can be used for non-credit cards.

Resolution 2008-01-9.1: For the card type, set length restriction of min=2, no max. Insert example of Canadian social insurance number.

Action 2008-01-9.3: John Boyer to perform spec edits and schema edits for card type change and example.

* ITS Rules Working group note: Any takers?

John Boyer: The I18N group wanted us to help define ITS rules for how to translate XForms documents. We said it might not be simple. Does anyone have a business need to work on XForms Translations.
Leigh Klotz: What's the ask? Produce examples? Meet to decide what to do?
John Boyer: They want someone to be a point of contact and become more familiar with ITS. Maybe the initial bit is to have a phone call.
Leigh Klotz: Is it this? http://www.w3.org/TR/its/
Steven Pemberton: What exactly do they want us to do?
John Boyer: They want to wind up with guidance to the world about how to translate XForms documents; either rules or guidelines for how to write rules for specific XForms documents.
Steven Pemberton: So it's a way of marking up things to be translated and not.
John Boyer: They have a v1.0 Rec but they may have a new set of requirements. The ref, for example, or value, in XForms.
Steven Pemberton: Or src.
John Boyer: Then how do you translate the document?

Leigh Klotz: I'll take a look at the document and I can participate in a call with the group.
John Boyer: The main thing is to contact Felix Sasaki; they're willing to do some work. They don't want to do the wrong thing with XForms.

Action 2008-01-9.4: Leigh Klotz to read http://www.w3.org/TR/its/ and contact Felix Sasaki to discuss XForms.

* Input Mode Issues

* Fix our shortname issue


John Boyer: Steven, I did want to wait until you got back to discuss this. There's been a bit of to and fro with Ian Jacobs on what the short names are supposed to mean.
Steven Pemberton: Did we decide on our philosophy?
Leigh Klotz: We asked for an xforms1 shortname and got a no.
John Boyer: No, it was xforms11 they didn't want.
Leigh Klotz: Like svg21?
Steven Pemberton: It does happen quite a lot; xhtml11.
John Boyer: So the question is what does xforms point to? Steven, you wanted the very latest document even if it's not a rec.
Steven Pemberton: In the future, when we have more than one major version, xforms should give you the most recent recommendation.
John Boyer: Good; that's what we want. For XML Events, it's pointing at the XML Events 2 Working Draft; that's caused some problems for XForms 1.0 TE, because it used the short name.
Steven Pemberton: I believe that normative references should use dated links since otherwise you spec changes under your feet.
John Boyer: So we have a bit of patchwork to do. It's a lot of work keeping up with the links as the editions change. Shortnames by version would be helpful.
Steven Pemberton: So XForms 10, a shortname for the latest edition of 10. If I said xforms1 I would expect the most recent spec in the xforms 1.* series, which is xforms 1.1.
John Boyer: Even if it's not a recommendation?
Steven Pemberton: Let's stick to recommendations for now.
John Boyer: I'm not sure if they agree with that. If they don't have any two-digit short names, how do you get to the latest working draft? So we need to ask W3 management to come up with a system.
Steven Pemberton: Did you get such a thing from Ian Jacobs?
John Boyer: He said when XForms 1.1 becomes a recommendation, that deprecates XForms 1.0.
Steven Pemberton: Here's the official W3C document on this: http://www.w3.org/2005/05/tr-versions written by Ian.
John Boyer: Steven, we're ok for now but eventually we'd like something better. Can you pick this up?

Action 2008-01-9.5: Steven Pemberton and John Boyer to continue discussing shortnames for recommendations and drafts.

* Proposal to let submission appear as an action and/or within submit

http://lists.w3.org/Archives/Public/public-forms/2007Dec/0041.html http://lists.w3.org/Archives/Public/public-forms/2007Dec/0042.html

John Boyer: Uli or Nick? We've been discussing strategies for not having a model element: submission or instance anywhere. We're still mulling through exactly when does the model element have to show up. Can I put a submission in the middle of an action sequence rather than calling a send? Do we want to add a position semantic to submission, so that based on where it is it might run without the usual mechanisms?
Leigh Klotz: If there's no model them where do you put definitions of submission that aren't run automatically?
John Boyer: (summarizes proposal)
Mark Birbeck: I think that would be confusing. But there's something I was going to propose; stick with the submission element, and the send action has the same attributes as submission, and then attributes on send override submission. So a use case might be a PUT and a GET to the same server. You could use a variable, but then you have to use setvalue. So you could do "send method=put". So submission is still a useful place to collect them attributes, event handlers, etc.
John Boyer: You may have proposed that ages ago as well.
Mark Birbeck: We didn't have a good way to do that.
Leigh Klotz: What about nested submissions, with just method overrides in your example?
Mark Birbeck: That could work; we need to look at the semantics of the actions.
John Boyer: We have a mechanism for doing it in 1.1.
Leigh Klotz: You can do both; you can have a nested submission and you can have the send with attributes. Then submission is always a definition and send is always an invocation.
Mark Birbeck: Send with attributes can work if you have no submission as well; it would be a "default" submission.
John Boyer: But that means you duplicate the submission attributes.
Leigh Klotz: You could do it with a schema type.
Mark Birbeck: Or put a send inside submission.
Charlie Wiecha: That's one of the proposals in Uli's note.
John Boyer: No it's submission inside submit. ev:event.
Mark Birbeck: That's another discussion. Allowing inline stuff; one of my guys suggested XML directly inside the insert action. You could have a send that can contain a submission that contains inline instance data. If you want to send a block of fixed data then rather than a ref you could have inline data. There's a general concept of having inline contents in addition to reference by id.
John Boyer: We're also using idrefs like this. We have the bind attribute. bind inside input would be funny.
Mark Birbeck: readonly without a model. It would be powerful.
John Boyer: I mean a bind element inside the input to identify the node to which the input binds.
Mark Birbeck: bind nodeset inside the input? I like that. For dynamically generated forms: data type, readonly, etc. All pretty handy. Otherwise you need two passes (form and model).
Leigh Klotz: Are we saying submission in send or duplicating the send attributes?
Mark Birbeck: Putting submission in send makes it easier.
Leigh Klotz: What about send method overrides?
Mark Birbeck: Hmmm.
Uli Lissé: I am thinking about send/submission. I don't like the Erik proposed of submission as an action. I do think replacing indirection by a child element is a nice way. I thought my proposal would give simpler authoring. nick: [irc] allowing submission attributes on the send is the same as allowing readonly on input
Uli Lissé: I don't think so. MIPs are related to a model item. Two or more form controls can be bound to a model item.
Nick van: You might want one readonly and the other not. It's the same concept as putting the submission attributes on the send. You just have to define how they behave.
John Boyer: Why do we want to get rid of the model element entirely? Right now submission is part of the model and we're discussing when we need the model element. One possibility is never.
Mark Birbeck: My motivation has been lazy authoring/onramp, the gradual stepping up. Having bind and submission outside the model is quite useful.
Nick van: I feel like that too. With a simple form and no need to do something when it succeeds or fails, there's no reason to have handlers in the model. If you need action handlers, it's good to add the model.
John Boyer: So there would be no events, or a hidden model?
Nick van: A hidden submission but no event handlers.
Leigh Klotz: Send or submit with attributes?
Uli Lissé: submission being a child of submit would override.
John Boyer: So you create a formal submission element as a child of send or submit. So we need a strawman proposal.
Uli Lissé: I don't like the attributes on send or submit. Where do xforms-submit-done go?
John Boyer: They go to the hidden implicit submission, which is pretty cool.
Nick van: I can try to write it up but I'm on holiday next week but I can start when I come back.
John Boyer: Does someone else want to do it sooner?
Mark Birbeck: Is this 1.2 or 2.0?
John Boyer: Right now it sounds like ease-of-authoring. We don't have enough clarity yet. But it could be 1.2. We have a short timeframe; last call this year.
Mark Birbeck: I wouldn't mind a go at it quicker than the next two weeks.

Action 2008-01-9.6: Nick van den Bleeken and Mark Birbeck to write up submission, submit, and send attributes and child nesting.

* Need collection of deferred issues from 1.1 in LC database

John Boyer: Can anyone do this?
Uli Lissé: I can do this.

Action 2008-01-9.7: Uli Lissé to collect deferred last call issues from http://htmlwg.mn.aptest.com/xforms-issues/ and put them into the Future Features area of the Wiki.

* Default instance when there is no model

John Boyer: We discussed this last year; is it the one without an id? Steven didn't like the change. For example, and XHTML page with a portlets, two of which contain XForms. Does the sudden existence of two portlets cause the entire document to change?
Leigh Klotz: XBL does something like this with anonymous IDs.
John Boyer: So you'd have to write it differently for use in portlets? That seems harsh. What about a containing form element?
Mark Birbeck: I'd like to see a form element with a collection of submission+instance in 1.2.
John Boyer: Each form element would set up a local context.
Mark Birbeck: Locally-scoped first-model, first-instance.
Nick van: [irc] I'm not sure we need instance outside the model, but if the model isn't there you have an implicit instance and when you add a model you could specify if there is an implicit instance, so the first instance isn't the default but the implicit one
John Boyer: We we concerned that if someone started with a form with no instance and then declared one we wanted to know if it was overriding the default instance. Then we decided that if there was an instance with an id then it was not overriding. Then what if it wasn't the first in build order; then as long as it doesn't have an id, it's an override, but Steven said that was a problem.
Mark Birbeck: Didn't we come up with some suggestion for the instance function to address either the first instance or the anonymous instance?
John Boyer: The parameterless one addresses the default, the first one in build order, whether or not it has an id.

John Boyer: We're about out of time. Let's have more discussion on email about this and pick it back up next week. Any volunteers to be the prime mover for this?

* Meeting Ends

* IRC Log