TPTP-PMC-20090218

Logistics

Attending: Joanna, Eugene, Paul, Oliver, Kathy

Any objections to last weeks summary?

• no

Does anyone need to promote agenda items?

• Paul has some timing constraints (in the car)
  • Wants to talk to 4.5.2 and 1.4.2 JVM support
• Eugene is ill and would like a brief meeting :-(

4.5.2 Finalization

• Paul started a test report before he left office
  • 89% attempt 2-3% fail rates
  • Some tests reporting as not yet run are are monitoring/etc items that should be trimmed from report
  • doctest results needs checkin
    • There was a question (Kathy) about why need to run these again
    • Joanna noted they were not 100% done before and you have to do 100% on final test pass of release anyway
    • In general, can do doc smoke tests on selected milesones though
    • Joanna and Kathy will sync on this
• There were trace.arm test question for Richard from Eugene
• Test project has no new issues. Ready to drop there
• Platform doc smoke tests more or less okay (some old defects and 1 new one that will be fixed in 4.5.3)

We had a formal vote on the following deferrals.

• Test
  • 239394 Cannot scroll to the bottom of the log file in log viewer events page
  • 173653 Issues with "Search and Replace" capability in Datapool Editor.
  • 237946 Can't delete Plugin-project completely some times in Test perspective
  • 243366 Copy and paste in encrypted datapool
  • 252958 Deleting multiple rows (records) from the datapool table corrupts the rows.
  • 253288 Unnecessary classpath entries deployed during test execution.
  • 256807 Revert function not working while deleting variables in datapool
  • 142045 Test Log Viewer: Overview page layout problems
  • 244362 Command Line Execution - overwrite = True flag does not actually overwrite the results
• Platform
  • 244331 - Profile failed with Unhandled exception
  • 254487 - Profile on server with memory analysis fails with segmentation fault error
  • 262975 - JVM Panic on JVMPI profile of simple class
  • 244203 - Impossible to run scheduler when workspace name contain accentuated characters (I.e cedilla) - CRITICAL
  • 247602 - [Martini] Class File instrumentation process is not thread-safe
  • 260290 - Profiled application VM crashes
  • 261451 - The downstream build will be failed if the remote connection is lost.
  • 115325 - 'Refresh' in Profiling Monitor view without saving lost active attach agent(s)
  • 143547 - New Agent Controller log file contains hard-coded paths.
  • 197512 - Profile As -> Eclipse Application; configuration and dev parameters encased in quotes
  • 235226 - [POG] Eclipse JVM out of heap space
  • 236359 - Processes run through remote Win AC/IAC intermittently lose last few lines of a terminated process's console data
  • 236424 - [Webdoc] [Readme] Vista extraction utility cannot be used to uncompress TPTP ZIP files.
  • 236887 - Support peer monitoring functionality on hosts with only an IPv6 address
  • 237707 - workbench cannot see a brand new secure ACServer running Sun JRE
  • 248302 - Incorrect "connection success" message with no IAC/AC running.
  • 253453 - Ramp Down Policy
  • 256207 - Recording is not happening in WID 6.1.2.000
  • 256227 - Plugin startup and NPE during shutdown
  • 258205 - ACStart.sh script not resolving symlinks
  • 258578 - ACServer not starting
  • 258812 - Add IBM 1.4 and Sun JREs to BVTs.
  • 258814 - Add BVT results to TPTP test reports.
  • 259133 - Provide support for Agent Controller and JVMTI Profiling on Linux 390 (Base implementation done. Need to still address Asaf's review comment)
  • 259186 - [POG] Confusion actions: Refresh vs Refresh view vs Unload Profiling Data
  • 259189 - [POG] View refresh should maintain tree list state
  • 262381 - Doc Smoke test failures.
• Trace
  • 252733 maj P1 sabihas@ca.ibm.com NEW Trace.UI All 4.5.2 Execution Statistics View is inconsistent
  • 126261 nor P2 sabihas@ca.ibm.com NEW Trace.UI Win 4.5.2 Execution statistic views do not save the column selection
  • 146947 nor P2 sabihas@ca.ibm.com NEW Trace.UI All 4.5.2 Execution Statistics view shows empty table instead of defaultPage.
  • 219379 nor P2 ewchan@ca.ibm.com NEW Trace.UI All 4.5.2 Allocation details tab opens even if "Track object allocation site" option isn't selected
  • 235914 maj P3 ewchan@ca.ibm.com NEW Trace.UI All 4.5.2 Trace data emitted via a Logging Agent is not loaded correctly.
• Unanimous acceptance

Paul wanted to mention special kudos for rampdon on 4.5.2

• Unanimous acceptance

1.4 JVM support

Paul notes that in 4.5.3, 1.4.2 breaks because of how Galileo works

• Could build with old EMF and run with old or new EMF
• Another component has this issue now (BIRT). Are verifying right now that can build on old and run on new
• There seem to be issues with this approach
  • RE would need to build for us that does Eclipse 3.5 with old BIRT, etc.
  • No guarantee that it will work even if they do it.
    • If project is backward compatible as it should be it will work.
    • If does not work then will go back to BIRT and ask for backward compatibility
    • We don't have too many dependences on BIRT so it would be a shame to lose 1.4.2 over this

Do not have a solid answer back from IBM stakeholders about whether 1.4 support is really needed? Kathy working on it.

• Two different questions
  • Do data collectors need to support 1.4 VM
  • Separate question from whether 1.4 is needed on workbench for the views
• Models running on AC and models running on workbench have to be consistent
• planning council said that release train does not require any specific JVMs
• Need to understand in the 4.5.3 timeframe. Any more discussion on 1.4 JVM

Build Servers

IBM mentioned that some new Eclipse policies about source bundling for builds (P2 repositories) makes it desirable to consolodate builds to one location.

• Wonders if Intel is interested in consolodating
• Chris mentioned that Intel had thought thru this a while back for solving some ofthe networking problems
  • Intel might be able to find some Itanium systems to donate to the cause
• Action item to Chris to follow up in Intel and close on whether this is feasible
• Would like to roll it out over next few milestones.
  • If possible would like to get done for M6 (sometime in march)
  • By then should at least have a plan

4.5.3 planning in general

Joanna put together 4.5.3 approximate schedule Our task now is to define P1 defects

• Cathy suggests moving all P1s targeted to Future to P2 then we can cleanly look at P1s targeted to our release
  • Targeted ones should be seriously triaged

Oliver would like to be conservative

• Test project was quite conservative before too. We need to be careful as we have new defects identified during process.
  • Paul notes he went in with 70% of time committed for 4.5.2 (30% flex for new stuff).
  • Oliver suggests maybe even as low as 50% for 4.5.3

UI and API changes needed by Mar 6 deadline

• Cathy (IBM), Chris (Intel), Eugene (Community) to probe their teams for P1 4.5.3 defect areas
• Need "need list" from consumers
• Map resources to create "can do" list
• Iterate with consumers to create P1 list.
  • Next week we hope to debate about what P1s can get done.

Misc

Eclipsecon

• Oliver said that room reserved
• Chris to borrow a projector.
• Intel to follow up on final attendees