Overview

Charlie: So there are a number of people due today, but I haven't heard from any except from the person from Spain

<Charlie> Agenda: http://lists.w3.org/Archives/Public/public-xg-app-backplane/2008May/0000.html

Charlie: do you know if Jack is coming?

Steven: No idea; I haven't seen him this week
... Is Mark coming?

Charlie: He hasn't joined yet

Steven: Well he spun off a new company called WebBackplane, so I guess he's interested?

John: [explains the background to Mark's company]
... He is travelling to San Jose at the moment

Charlie: I pinged Scott McGlashan of Voice asking if he couldn't come to send someone else from voice
... but I didn't get a reply in the time
... so here are three of us in any case
... the agenda lists the topics and some documents

[1] Panel at the 2006 Tech Plenary:

http://www.w3.org/2006/03/01-TechPlenAgenda.html

[2] W3C Note on the Backplane: http://www.w3.org/TR/backplane/

[3] Amsterdam meeting minutes:

http://www.w3.org/MarkUp/CoordGroup/2006/backplane-meeting.html

[4] Amsterdam meeting report:

http://lists.w3.org/Archives/Public/www-tag/2007Jan/0035.html

[5] Backplane XG Charter:

http://www.w3.org/2005/Incubator/app-backplane/charter-20080409.html

Events

Charlie: Looking at the minutes from the meeting, I think the eventing issue is really the place to start

http://lists.w3.org/Archives/Public/www-tag/2007Jan/0035.html

Charlie: data and controls should be signalling what they are doing
... and if we can set expectations about when that happens, then it can make composition more transparent

John: When we first presented the backplane, I said that XML events is the nexus of interoperability
... so I agree
... but I haven't developed that since then
... but I do now have experience with using XML Events for combining stuff (such as XML Signature)

Charlie: I have a tax application that links in to the MVC structure using events without destroying the principle

John: The new upcoming draft of XML Events makes it easier too, since it has conditional execution

Charlie: What this XG can do is promote some patterns around notification events
... existing mashups in Javascript do not make the model transparent

John: More value for event context could come from handlers with named parameters

<John_Boyer> For example, this way a module could indicate whether it had take some action related to the event, which other modules could use as the event bubbles up.

Steven: One thing I would like to see is that we find a unified treatment of what in XForms is called relevance
... where parts of the tree become irrelevant to events, and so must not be used for setting off handlers
... several of our groups have such a need, and everyone does it differently

Charlie: Another apparent point of disagreement but is a point of alignment is that subsetting of events is OK

Charlie: About adding value up and down the bubble capture
... can we recast aspect-oriented programming into an event-based model?

<Zakim> John_Boyer, you wanted to comment on orthogonality of relevance and conditionality and to comment on event profiling e.g. to target only and to comment on importance of event

John: On the bubbling part, I agree
... it is the capture and bubble phase that makes eventing a good way to get modules to talk to each other
... for instance I have been doing stuff with ODF + XForms 1.1
... ODF can listen for the bubbling of an XForms serialise event,
... and look at the target
... and then examine the submission element to look for ODF serialization mime types
... XForms by itself could not have specified that
... because different host languages would be different.
... About profiling XML Events
... there are some who want to do something similar, but for instance only process at the target
... if you look at XForms 1.1 conformance, we say you MUST implement ev:target, but allow profiling too
... A host language may not want to implement everything
... Now that is a mixed message that I have just given
... but I think we need to strike a balance
... Back to Steven's point about relevance
... this notion is orthogonal to merely being able to create new handlers

Steven: XForms relevance and SMIL timing both have the same effect, to remove a bit of the tree from processing for the time being (albeit different reasons)
... but the concept is the same

John: You are saying that as a cross-cutting concern, rather than using if statements to exclude bits of the tree, you can use some declarative conditional to exclude them
... There are some things that are so hard to say in programming, that aspect style is the only way to do it

Steven: I think the problem is that DOM Events was designed based on HTML which doesn't have relevance (as we call it)
... but then along came several specs that discovered the concept independently, and what we need to do is fold it back into the centre

John: In a modularised way so that we can do this incrementally

Charlie: And I think we need to identify some event types that all applications should be able to expect

Steven: I agree, such as the load event (document is ready)
... so that an app knows when everything in the combination is ready to go
... Submission is another one, such as John's case above
... so that each application subpart doesn't need to invent these events themselves

John: We have a techplen coming up, and they always ask for talks

Charlie: Well, we need a definitive report on needs for this area

John: We should be doing the requirements work, not necessarily the technical work

Steven: An XG is only for the groundwork, not the technical work

How Further?

Charlie: Maybe I should ping people by private email

Steven: I haven't got into the habit of checking this list yet, so that would be a good idea

John: I didn't get any list mail yet

Charlie: Can I check the list membership?

http://cgi.w3.org/member-bin/list_mgnt.pl

Steven: Only 4 on our list

Charlie: So that's the problem! I assumed if you joined the group you would automatically be added to the list.

Steven: I should point out that I am here purely in CWI mode, since XG's do not have any staff resources

Charlie: So let's update the list now

<John_Boyer> http://www.w3.org/2000/09/dbwg/details?group=42340

<John_Boyer> Participants link in this: http://www.w3.org/2005/Incubator/app-backplane/

Charlie: Shall we do the next meeting in 2 weeks?
... but do we in general want to do weekly or bi-weekly?

Steven: Weekly at this time would clash with HCG.

John: So which day would we do it?

Steven: I only have Monday at this time left. All other days are taken. This hour is teleconference rush hour.

John: Groan, Monday early in the morning... let's not do that

Charlie: OK, two weeks from today then, Friday 30th May at 14:00UTC, and we'll talk more
... thanks for turning up, and talk to you soon.

[ADJOURN]