Skip to main content

PMC Conference Call - March 22, 2005

WTP PMC Minutes for March 22, 2005 Conference Call

Attending: Tim Wagner [TW], Arthur Ryman [AR], David Williams [DW], Dominique De Vito [DV], Jochen Krause [JK], Naci Dai [ND], Bjorn Freeman-Benson [BB], Christophe Ney [CN]

Others: Lawrence Mandel (representing WTP community) [LM]

Procedural matters

  • TW: Reminder that private mailing list ( has been decommissioned and all PMC members and other interested parties should subscribe to the Eclipse-hosted one (
  • AR: Requested that EMO investigate IRC server for telecon use.

Community [LM]

  • OOPSLA submissions for tutorial and panel completed
  • JK and DV also presenting at conferences; will update Lawrence with info
  • Messaging: no response yet from WTP community; LM requested that PMC members appeal to their respective organizations for support in writing articles and communicating WTP info; EclipseCorner in particular is easy to publish to and eager for material
  • TW: BEA eWorld, dev2dev interview, and JavaOne conference are likely candidates for WTP presentations; more as the event dates/agendas crystallize
  • CN: ObjectWeb hosting conference in Viet Nam at which WTP information will be presented
  • WTP help system initial content should be available by next week (delayed from previous estimate); currently working towards stable local builds

Documentation Strategy

  • PMC discussed the likelihood of a lag between code GA and documentation readiness, especially in light of the limited resources and the size of the initial documentation content drop from IBM. If we ship in July, a September doc release, with subsequent quarterly updates, might be a tractable schedule. Expect to use professional writers (devs may author initial content or edit subsequently). Initial code release will contain samples, JavaDocs, and anything else thats ready at that point.
  • Documentation will have a separate build process/schedule. LM/AR to pursue offline.
  • AR: IBM has product documentation resources that will contribute to this effort
  • TW: BEA probably in similar state; some fraction of a doc resource likely available to help post (BEA’s) 9.0 delivery
[PMC members only for the remaining portion of the meeting]

Requirements Group Report [JK]

  • Requirements group composition:
    • Paul Meijer (BEA)
    • Todd Williams (Genuitec)
    • Arthur Ryman (IBM)
    • Jochen Krause (Innoopract, chair)
    • Ivelin Ivanov (JBoss)
  • First meeting will follow next week’s PMC call
  • Version numbering – PMC discussed merits of next version being “1.1” (historical consistency, description of maturity of project) versus “3.2” (alignment with platform, less confusion for users). Tabled final decision pending feedback from Eclipse Councils next month.

Architecture Group Report [DW]

  • Several new bugs posted [see query string in last week’s minutes –tw]
  • Thursday call agenda: Browser resolution (move to platform?), tabbed property page (same question)
  • TW: What role is Architecture Group playing in API review? DW: Focusing on WTP vs. platform and cross component issues as detected by themselves, component leads, or members of the community.
  • TW: Suggested that a community review of the APIs in the form of 1-2 hour presentations by each component lead would be a forum for API commentary. Architecture Group could help structure the review, but the community at large would own the outcome. AR noted that timeframe must be post M4 to allow API definition process to complete.

Disk Quota/Build Issues [AR]

  • Problem is not so much disk quota on our main server as the load we place on mirrors; if we trim down the nightly size, that will have the biggest impact
  • Three suggestions
    1. Don’t retain nightly builds beyond the next integration build
    2. Don’t publish non-SDK bundle in nightlies (they’re only for dev use)
    3. Don’t publish WST-only bundle in nightlies (same reason)
    PMC resolved to adopt all three changes to the process, and that “warmup” (pre-final) I-builds need not be kept after the next successful I-build.

Integration Builds [ND]

  • Platform insists that test failure implies the result is not an I-build; we’ve lacked that level of quality in our I-builds
  • Our previous milestone releases have been stressful because of accumulated test failure debt
  • Current (updated) process is to attempt the I-build on Tuesday, focus on build or test failures midweek attempting to produce a successful build on Thursday. Expectation is that there are no large changes between Tuesday and Thursday; changes can be released to head, just not included in the I-build candidate.
  • Biggest remaining problem is getting people to focus on tests; will now require tests to pass to release the I-build (otherwise we fail for the week and try again the following week rather than posting the failure).
  • [AR] For m4 we have two buckets: “real” tests, and API tests that do not yet have backing implementation and will therefore fail. For the duration of m4, we need to manually differentiate in declaring the I-build green.
  • [ND] Also, Linux machine should be up this weekend, so we will start having parallel builds on Linux as well as Windows.

API Review

  • PMC reviewed initial API statistics; will track component convergence over time to help drive process.

RDB / DTP Issues [AR]

Arthur posted minutes of previous meeting to mailing list. Components would likely move in August (pending proposal approval); still considering whether to rename them in advance to limit later churn on other WTP components that are downstream.

Action Items:

  • [JK/DV/CN] update Lawrence with info on upcoming conference presentations of WTP info
  • [TW/DW] work on scheduling API review by component leads
  • [ND] update build scripts based on the “Quota” resolutions (see above)
  • [ND] update build page with process change (test pass requirement) for I-builds
  • [AR] ask Jeffrey to add time series tracking to API statistics pages so we can track trends at the component level
Minutes taken by Tim Wagner, March 22, 2005

Back to the top