Interoperability Minutes 2007-06-06

From MemberWiki

Jump to: navigation, search

Full minutes: http://www.openajax.org/member/wiki/Interoperability_Minutes_2007-06-06

Contents

OpenAjax Alliance Interoperability Committee meeting minutes 2007-06-06

Attendees

  • Kin Blas <jblas(at)adobe.com>
  • Gideon Lee <glee(at)openspot.com>
  • Jon Ferraiolo <jferrai(at)us.ibm.com>
  • Ted Thibodeau <tthibodeau(at)openlinksw.com>
  • Bertrand Le Roy <Bertrand.Le.Roy(at)microsoft.com>
  • Greg Murray <greg.murray(at)sun.com>
  • Rama Gurram <rama.gurram(at)sap.com>

Original Agenda

  • Agenda
    • Plan for finishing off OpenAjax Hub 1.0
      • One final open issue: Issue 14: Should pub/sub callbacks be passed the list of tokens?
        • We started talking about this at our last meeting, but didn't resolve the issue
      • Tasks not yet completed
        • (Re-)implement refOrName as a string (with late-binding name resolution)
        • Add tests for scope!=window (and refOrName as a string)
        • Finish polish work on the spec
      • Optional tasks
        • Make sure test suite is suitable for other Hub implementations?
        • Make sure test suite works outside of JSUnit?
        • More work on shrinking the reference implementation? (currently at 2845 bytes after Dojo ShrinkSafe)
        • Provide optional JavaScript for parameter validation?
      • Ensuring that the Hub meets industry requirements
        • Any volunteers to put current implementation into practice, particularly in a mashup application?
          • Jon will use Hub within Ajax Product Selector application (for Mktg WG)
        • Are we convinced the APIs are sufficiently future-proof?
          • Versus requirements of upcoming Communications Hub?
          • Versus Crockford's proposed <module> tag (http://json.org/module.html)
          • Versus Google Gear's WorkerPool?
          • Should we agree on a small number of standard OpenAjax events (event name and payload) before we release Hub 1.0? (For example, a DataChanged event on a UI control.)
    • OpenAjax Registry

Minutes

Topic: Plan for finishing off OpenAjax Hub 1.0

One final open issue: Issue 14: Should pub/sub callbacks be passed the list of tokens?

Jon: Bruce asked whether we should add a 4th parameter which contains the list of tokens after splitting the event name, which almost certainly the Hub will have to do anyway.

Kin: Didn't we discuss this last time.

Jon: Yes, but we didn't resolve this issue before the meeting ended.

Kin: I don't see the use for the extra parameter. Just as easy to call split if you need to. Often you will have a case statement that uses the string value for the event name.

Jon: Others said the same thing last week.

Jon: Any other comments?

(silence)

Jon: OK, I'll close ISSUE 14 and send an email that highlights the fact that the issue is closed, especially since Bruce brought it up and he isn't on the call today.

Tasks not yet completed

(Jon talks about the list of tasks that still need to be completed)

  • (Re-)implement refOrName as a string (with late-binding name resolution)
  • Add tests for scope!=window (and refOrName as a string)
  • Finish polish work on the spec

Jon: refOrName was implemented in a previous version of the Hub. Probably another four lines of JavaScript.

Jon: Anyone care to volunteer to help?

(silence)

Jon: No problem. I can finish those tasks. Most are short. The hardest task will be the wordsmithing tasks on the spec.

Optional tasks

(Jon talks about some highly desirable but not required tasks that we might pursue on Hub 1.0)

  • Make sure test suite is suitable for other Hub implementations?
  • Make sure test suite works outside of JSUnit?
  • More work on shrinking the reference implementation? (currently at 2845 bytes after Dojo ShrinkSafe)
  • Provide optional JavaScript for parameter validation?

Jon: The first item above, making the test suite suitable for other implementations, has been requested by some members. I might work one-on-one with members who have requested this to do our best to meet their needs.

Jon: Howard submitted code for open source project where the tests will work without JSUnit. I have commented that out, but I was hoping to resurrect it. This might help with the first bullet.

Jon: Regarding further shrinkage, we might be able to cut out a few bytes by using shorter local variable names. I have heard that keeping the Hub less than 3K would tend to make toolkit vendors include it without question, so let's try to make sure we stay under that size.

Jon: Does anyone want to comment on the above optional tasks or volunteer to help?

(silence)

Ensuring that the Hub meets industry requirements

Jon: We know that the Hub is in the ballpark for meeting industry requirements because of the InteropFest and the 5-6 companies that used the pub/sub mechanism during that event. But I wanted to make sure that we felt confident that we were shipping something that met the community's needs.

Any volunteers to put current implementation into practice, particularly in a mashup application?

Jon: It would be best if there were trial applications that used the current version of the Hub. I will be using it in the Ajax Product Selector application that we are developing in the Mktg WG. Anyone else planning to use the Hub in one of their applications?

(silence)

Jon: OK. I'll try to recruit TIBCO offline since they have done similar demos in the past. Also, I'll try to get a mashup application to use the Hub, starting with IBM's QEDwiki team.

Are we convinced the APIs are sufficiently future-proof?

Jon: I also want to make sure we do a sufficient paper analysis of the Hub to make sure it fits in with Ajax requirements that are emerging from other technology areas. Here is a list:

  • Versus requirements of upcoming Communications Hub?
  • Versus Crockford's proposed <module> tag (http://json.org/module.html)
  • Versus Google Gear's WorkerPool?
  • Should we agree on a small number of standard OpenAjax events (event name and payload) before we release Hub 1.0? (For example, a DataChanged event on a UI control.)

(Jon enters URL for Greg Wilkin's bridging proposal into IRC: http://www.openajax.org/member/wiki/Communications_Events_Proposal)

Jon: I realize this is all coming out of the blue. What I suggest is that we talk about these future proof issues at the next phone call in two weeks and attempt to decide whether we are satisfied that we are sufficiently future proof at that time.

Jon: I see the potential need to bridge to servers, to other mashup components, and to local storage. Anyone read Greg Wilkin's proposal yet?

Kin: Briefly, but I need to think about it.

Jon: I will attempt to get an email discussion thread going about bridging.

Jon: Is anyone aware of Crockford's module proposal?

Rama: Yes

Greg: Yes, but haven't looked at it recently

Bertrand: Is it getting traction? Will it be part of HTML5?

Jon: I don't know for sure. Based on discussion at yahoogroups caplet list, I would guess that it is in the same stage it was in last fall, where it is an interesting proposal without any specific positive or negative decisions by the browser vendors, but I'll try to get an update from Crockford.

Jon: How about the need for some standard events and payloads before finalizing the Hub?

Bertrand: We already have a couple of standard events. I would say no.

Greg: We have some conventions in this area with jMaki for onSave and onSelect. Maybe there could be one for onLoad. We have a table of suggested events and associated payloads. Within jMaki glue. It's a short list.

Topic: OpenAjax Registry

Jon: This is the first meeting when we will formally discuss how to proceed with the Registry. We have had lots of preliminary discussions in the past, but we are now discussing it formally and it's time to make decisions. Let's start by walking through my strawman proposal. I want to emphasize that this is just a first draft strawman. I figure that we we can walk through the strawman proposal and will discover various issues and implicit requirements as we make progress.

Jon: First topic from proposal: register location and format. I am proposing JSON format with a particular URL on our Web site.

Bertrand: Should it be JavaScript instead of JSON file so that it can be integrated into applications?

Jon: Do you mean calling the file *.js instead of *.json, but requiring that the file be in JSON format?

Bertrand: No, that the file is *.js that creates a registry JavaScript object.

Jon: Makes sense to me.

(some discussion about how our globals registration does indeed include fooobject.prototype registrations)

Bertrand: What about expandos on DOM elements? For example, Prototype adds other methods and fields to DOM elements. Potential for collision there, also. Other browsers than IE use DOMElement for this. We could use that convention for the registry. Just need to pick something.

Jon: Something to look at. Personally, this is new to me and I have to do some homework.

Jon: Let's talk about my process proposals.

Bertrand: Make sure that the wiki page informs the public that they need to send email to public@openajax.org.

Jon: I agree.

(Jon summarizes process for discussion via phone calls, wiki, and email.) I was trying to propose a process with sufficient rigor but as simple, fast as possible leveraging electronic means as much as possible. Anyone want to comment on whether this is good or bad?

Ted: As appropriate as anything.

Jon: Last section is on versioning, where I propose that registrations are forever. I don't see how else to do it since old libraries will be in use forever. Therefore, it doesn't make sense to remove a registration, even if the given toolkit quits using a particular global.

Jon: OK, out of time. We'll look at Microsoft's list of globals in MS Ajax to see if we can start to glean categorizations and criteria for registry entries.

Retrieved from "http://www.openajax.org/member/wiki/Interoperability_Minutes_2007-06-06"
Personal tools