Interoperability Minutes 2007-04-25

From MemberWiki

Jump to: navigation, search

Full minutes: /member/wiki/Interoperability_Minutes_2007-04-25

Contents

[edit]

OpenAjax Alliance Interoperability Committee meeting minutes 2007-04-25

[edit]

Attendees

  • Phil Berkland <berkland(at)us.ibm.com>
  • Kin Blas <jblas(at)adobe.com>
  • Gideon Lee <glee(at)openspot.com>
  • Jon Ferraiolo <jferrai(at)us.ibm.com>
  • Kevin Hakman <khakman(at)tibco.com>
  • Ted Thibodeau <tthibodeau(at)openlinksw.com>
  • Bertrand Le Roy <Bertrand.Le.Roy(at)microsoft.com>
  • Bruce Johnson <bruce(at)google.com>
  • Rama Gurram <rama.gurram(at)sap.com>
  • Ingo Muschenetz <ingo(at)aptana.com>
  • Greg Murray <greg.murray(at)sun.com>
  • Adam Peller <apeller(at)us.ibm.com> (joined the call late)
[edit]

Original Agenda

  • Agenda
    • Interoperability Working Group
      • Interoperability Working Group Creation Review phone call
        • Thursday, May 3, 2007, 8amPST, 11amEST, 5pmParis
        • US toll-free: 1-888-227-9220, Int'l: 719-785-4480, Passcode 871757
      • Member voting - May 4-May 11, 2007
      • Steering Committee voting - May 11-May 18, 2007
    • Hub Issues
      • Issue 2: Pub/sub support for hierarchical event names and wildcards
      • Issue 3: What to do about the 'load' event and initialization management
      • Issue 4: What's our recommended best practices and/or standards around how to name events
      • Issue 5: Event versioning
      • Issue 6: Are events specified by a pair {prefix,EventName} or just by an EventName
      • Issue 7: Should the Hub's own events be "/OpenAjax/foo" or "/OpenAjax/Hub/foo"?
    • OpenAjax Registry
[edit]

Minutes

[edit]

General

  • Reminder re: chartering process for the interop WG; creation review phone call (open to everybody, not just Interop) next Thursday at 11am EST. Opportunity to hear about and express opinions about the charter before we vote.
  • Zip file attachments are problematic for some companies, so please send using something other than a zip in email.
[edit]

Issue #2: pub/sub support for hierarchical events and wildcards

  • Jon, Howard, and Bruce have said in email that they are in favor of #4
  • Review of basic ideas:
    • Hierarchical string event names, separated by token separated (e.g. '.', '/')
    • Should we be able to have wildcards? And if so, in which token slots?
    • Howard has submitted an implementation of #4 that shows it is a small number of lines
  • Bertrand: I was originally in favor of the simplest approach that works, but I believe that #4 can be implemented efficiently and agree with it.
  • Kin: Favor #4 also. To clarify, ** at the end matches multiple levels? (Jon: Yes)
  • Ingol: Favor #4, assuming that we're only talking about '*' as being special, not indicating a regular expression.
    • Jon: Right, "*" is not like a regex. But you can still do regexs inside the filter function.
  • Bruce: I was sold because it makes the easiest thing for the developer the most efficient choice. Its absence might encourage app-level code to be net bigger, slower.
  • Greg: Have tried several approaches along these lines, and regexs are much slower.
  • Rama: Also like #4.
  • Gideon: Proposal #4 seems just fine. No strong opinion.
[edit]

Issue #3: managing library loading

  • At the FTF, the hub explicitly got involved with loading, but maybe that's too heavyweight
  • Bertrand: Why not just wait
  • Question (unidentified person): What does "ready" mean?
    • The toolkit code has been parsed and any initialization has been completed
  • Bruce: How do toolkits load? How does it interact with <script> tags?
    • At the FTF, we saw that there is a wide variety of init choices
  • Kin: Still fuzzy as to what "initialized" means. Does it mean that (a) init = all globals and function parsed or (b) all elements have been processed and the page is ready for interactivity
    • Bertrand: It seems like (b) to me. There may be a third option, where there are different phases.
    • Kin: Our JS files are self-contained, so that there is no need for users to worry about interdependencies. They are mix & match, so there is no central point.
  • Jon: Maybe we should just drop this whole area right now. We don't know what to do right now, and it doesn't hurt to wait.
    • Bruce: I agree. If we're wrong, people will get turned off.
    • Kin: Are all the use cases written down on the site?
    • Jon: No, not yet.
    • Kin: Let's get everyone who represents a toolkit to describe their loading mechanism in the wiki so that we can design around that.
    • Jon: Good idea, but let's wait until after 1.1 before emphasizing this.
    • Job: Any objection to dropping load features from 1.0?
      • (no objections)
[edit]

Issue #4: Best practices about event names

  • Jon: Separator character? Folks on the email list seemed to prefer dots.
    • Ted: Unless there's an argument in favor of '/', let's go with dot.
    • Jon: Anybody object to dot?
      • Kin: Is there ever going to be a need to see if something is at the root? (e.g. xpath's "/*")
      • Jon: We agreed over email that we only want absolute names
      • Bruce: Event names are not strictly-speaking hierarchical anyway, so dots seem to imply a hierarchy less (a good thing).
  • Jon: Are we going to only allow domain-names as event-name prefixes, or do we want shortcuts?
    • Ted: Administering registry for shortcut names seems like a nightmare
    •  ???: Could we leverage our Globals registry?
    • Ted: I'm concerned about the Global registry itself as a maintenance headache
    • Jon: My inclination is to go with domain names; the simplicity and regularity of the mechanism trumps special-case shortcuts.
    • Bruce: Agreed with using domain names. We shouldn't give OpenAjax a special shortcut name, though.
      • (murmurs of agreement)
    • Jon: Any objections?
      • (no objections)
[edit]

Issue #5: Event versioning

  • Jon: Contracts can change over time. How can we help (if we choose to) with the issue of versioning event payloads across time?
  • (Jon reiterated the choices from the email)
  • Bertrand: I prefer the last one: make the payload itself include the version info; can be added after-the-fact if needed.
  • Kin: Agreed that it should only be part of the payload.
  • Jon: We have to think about how to encourage best practices, including best practices.
  • Bruce: I think we should consider a recommendation on including or not-including version info as part of the event name. IOW, is that a good practice or not?
  • Jon: Cited the W3C being unable to settle on a best practice. We could
  • Ted: Best practices can evolve, so maybe we should actually say that we don't recommend it if that's what we conclude. If we find it useful over time, then we can update the best practice.
  • Jon: Let's make this decision after we have worked out our own events.
  • Jon: Anybody disagree with assuming the version info is event-specific and in the payload?
    • (no objections)

(ran out of time before finishing all issues)

Two weeks from now is JavaOne. Several of us will attend, so next phone call in three weeks.

Retrieved from "http://www.openajaxalliance.org/member/wiki/Interoperability_Minutes_2007-04-25"
Personal tools