Skip to main content
Jump to: navigation, search

TPTP-PMC-20080430

Logistics

  • Attending: Joanna, Chris, AlexA, Paul, Oliver
  • Regrets: AlexN, Harm

Are there any problem with minutes from last time?

  • No.

Ganymede

I7 Testing

Oliver asks what the status of testing is?

Paul states that Test has a few regressions, a few blocking issues.

  • Team is working thru those items
  • Test pass nearly complete despite issues
    • A bit of UI layer refactoring is still in process
    • Functional testing there has proceeded manually until UI works
  • Team did not have cycles for full test pass
    • TP2 is a smoke test pass.
  • Some of the regressions will likely spill into I8
    • The biggest issues was the platform defect with IAC that is being fixed now.

Joanna states that Platform will just have time to finish test pass 1

  • Oliver asks if Joanna be okay with what will be done by the end of TP1?
    • Joanna will be reasonably okay.
    • A lot of work in this iteration that IBM has been around resolving criticalish defects identified by stressing from IBM's TPTP derivitive works
    • There was also some Translation testing in this iteration
  • Joanna states that Platform will finish the test pass but it won't be quite as full as originally desired

AlexA states for Trace/Profiling that team is also in test pass

  • Concentrating on some core bits
    • BTM: all tests run and works okay
    • Only 1 blocker stopping team from testing on linux platform

AlexA notes that there has not been a build lately.

  • Team is currently testing with custom built binaries
  • Joanna asks whats up with build
    • AlexA states that there is an issue on IBM side of build
    • Joel to look into IBM side right away
  • Kiryl will be checking in on builds a bit during Russia holiday

Oliver asks if there is anything else regarding the test pass that needs resolution in PMC meeting.

  • No

Ganymede I8 Discussion

From here to end of release, we want to be concentrating on stabilization and testing as primary objectives.

Paul suggests that for I8 development period (next week) that project lead approval for defects. Following that, move to PMC approval only for rest of release.

  • Oliver thinks leads should be in control of their teams over what gets fixed during that 1 week of development
  • Oliver asks if this is simply stressing our normal behavior or if this is different.
  • This is slightly different (a bit more conservative) but quite similar
    • Paul expects a rush of defects and would like to have some extra eyes reviewing to calm the tide a bit.
    • For Test project, Paul does commits/finalization for most of the stuff on behalf of a number of contributors.

Joanna is mostly worried about the regressions that may pop up from here to the end.

  • Joanna expects that we should be more or less in a continuous test pass between now and the release.

Oliver suggests that leads should be working on list of defects that they intend to fix in I8.

  • Paul suggests that leads should be ready to talk to this by next week PMC.
  • AI to leads to come with I8 lists next week.

Oliver asks if there are any other items related to Ganymede development? AlexA raised a possible Trace issues regarding whether or not Richard will be able to finish his items.

  • Updating key components from old to new execution framework: this and this.
  • Richard may not have time to complete in remaining time
    • AlexA asks if someone available to help Richard? e.g., maybe Bing or Eugene?
    • Joanna notes that late yesterday Jonathan sent long answer to Richard's platform questions (on platform dev list)
      • Richard thanked him for the detail. As far as Joanna knows, this clears up any blocking concerns that Richard has.
      • AlexA notes that even if questions are answered, the engineering of task may need help as well and feature is at risk.
      • Joanna notes that if there are more specific questions/needs from Richard then Jonathan/bing can help but it will probably be in the form of consulting on questions
      • AlexA will talk with Richard tommorrow (during Russian holiday). If there are still concerns about risk of feature, please copy Oliver on mail thread.

Ganymede Release Items

Paul noted that mail is starting to go around regarding release reviews for Ganymede

  • Its looking like they will be June 4 or June 11
  • Oliver asks about effort involved
    • Typically we take previous year and freshen
    • Did paul ever find review documents from last year?
      • AI to Paul to try to find presentation from last year for Oliver
  • Chris noted that the release document also turned out to be a little timesink last time.
  • AI leads to go thru presentation (when Paul finds it) and documentation of rampdown policy
    • There is definately a question around the section on enhancements that will not be completed.

TPTP (mainly profiling) and IBM JVM

There remain a number of issues (mainly JVMTI border conditions) that have shown differences in behavior between the IBM JRE and the reference implementation. The has been discussed in a variety of bugzillas.

  • Harm would like to drive a discussion at IBM around closing these out.
    • Harm would like to drive for a clearly defined interface between IBM VM team and TPTP folks to track these bugzillas.
  • TPTP should finish collecting all these issues into a single enumeration to facilitate this discussion.
    • Question: Is the list of defects that AlexA sent out the full list of discussion topics?
      • Joanna notes another blocker with Java6 surfaced recently with an application surfaced recently
      • Some odd java -version behavior
    • AI Joanna to forward "final" list of bugs including the new one
    • AI Chris to include in minutes so that we can point interested parties at the minutes.

SAP Discussion

Oliver asks about the status of understanding SAP functionality and integration opportunities

  • AlexA had another call with Andreas
  • discussed scenarios that might be able to integrate
  • From Alex' discussion, they like the idea of simply using extension points to plug in but not necessarily having widespread dependences/integration to rest of project
    • Oliver noted that this would have the feel of being a consumer of TPTP
    • SAP is not going to have much time to do a lot of integration work
    • We(Oliver?) may want to get clarification from them about how much integration work they would take on.
  • Main desire from SAP at this point is around getting details aroun dextension points
    • This consumption of allocation point information would definately require some development to create appropriate extension points
    • It would require some more interfaces to VM (possibly tweaked instrumentation hooks?)

Oliver asks how long AlexA thinks it would take to implement the extension points based on discussion about Andreas' wish list.

  • Alex' gut feal is that it is about 2-3 person months of effort
  • Staffing at that level is a question mark.

More discussion around this at AG call on Monday

  • Understand a bit more about expected man months of effort
  • Oliver thinks that for project success finding a way to get contributors to TPTP might be good (even as a trade... e.g., might it make sense for TPTP to find 2-3 engineer-months and get a commit of similar size from SAP in another area?).
  • Eugene was going to kick the tires of the static heap analyzer
    • AI to Joanna to be sure Eugene should provide his feedback by AG call.

MISC

Alex notes that Russia has a holiday thru end of this week but will be working on Sunday.

Oliver asks if there is any update on Intel resourcing/transition plans

  • No update

IPZILLA Java6 StackMap BCI Feature

Chris asks about folks experiences with IPzilla

  • Apache Harmony team is closing on their milestone 6 that fixes some verifier gaps that prevented our consuming the technology for populating stackmap attribute on intrumentation of Java6 classes.
  • Code is freezing quite soon and will be released in May.
  • Plan to get IPzilla review after code freeze
  • Some minor risk that Harmony may find a bug unrelated to our use and change the relevant code between the frozen version that is reviewed and final release.
    • Wanted to get peoples opinion of theoretically going with a pre-version Harmony Milestone 6 that is reviewed.
    • Paul noted that there used to be a ~250 line threshold allowed for changes (particularly related to 3rd party contributions)
    • General thought is that if IPzilla approves Harmony milestone 6 based on frozen code review and then <~250 lines change we should be good.

Upcoming Meeting Agenda

Next Week PMC

  • What will be deferred (i.e., fall out of 4.5?)
  • Oliver trying to get Harm to join.

AG call

  • I8 priorities
  • IBM constraints on moving features out of 4.5
  • SAP

Back to the top