TPTP-PMC-20090909

Logistics

Attending: Paul, Chris, Eugene, Kathy,

• Ernest has meeting conflict
• Oliver ran long in another meeting

4.5.2.1

• M3 is pretty much closed
• There are new requests for backfixes for a few more items in M4
• No schedule yet. Kathy needs to coordinate with consumer to find a desired date
  • She will probably request a schedule in next few weeks
• We discussed overlaps between 4.6.1, 4.6.2 and retrofitting some bugfixes to 4.5.2.1.
  • There is a belief that there are NO overlap issues
  • 4.6.1 is finishing now, 4.5.2.1 M4 will be done during development of 4.6.2 (and it features 4.6.2 targeted defects anyway)
  • Currently it is looking like there will be a request for about 6 (1 build, 1 test, 4 platform)

4.6.1

We are in a test pass now.

• We saw about 1 regression from Test and about 3 from Platform
• We were dropping RC3 candidate today
• Will will create a final candidate this Friday and dropping early next week (midweek)
  • We won't have time for a full smoke test pass (resourcing limits)
  • Paul would like to agree on strategy for test pass because of resourcing
  • Paul wants to try key use cases but doesn't have time for much.

Paul asked if we want to do a full test pass report. or just suffice with a "pass/fail" indicator for each project.

• Kathy is okay with pass fail. General agreement that there is no reason to butcher the test report for such a light thing
• We discussed that in future probably need to fine tune how smoke testing works (it has grown to be pretty rich over time) but need it to be a bit more scalable to available resourcing or time.

Paul asked about ongoing issues with builds. Eugene thinks the issues are resolved now even though the All-in-one must be rebuilt manually.

• Issues were caused by need to change the build scripts at both IBM and Intel

Kathy noted that there are a few (5) P1s to be deferred

• She will be sending list out onto list.
• Three are issues that Intel China is looking in. The team got caught up in a priority blocker that took more time than we thought so it is unlikely that these three will get done in time.
  • These 3 will be P1 for 4.6.2.
  • Paul asks if Intel has the resources available for 4.6.2. Chris noted that Intel is continuing with current efforting during winter maintenance
• One issue is from Eugene about freeing memory.
  • The suggestion is to move this one to a P2 for 4.6.2
• One relates to integrating the agent controller with some other parts of the build.
  • This one will probably go down to P2 in 4.6.2

Intel has 1 more patch (trying to isolate the memory problem), Jonathan has a patch by about end of week or next week. Worst case Friday or over the weekend. Might have to do a quick stress test on stuff done over the weekend.

4.6.2

Will discussed whether we will continue to build on the 3.5 maintenance stream or if we will we jump to 3.6 stream?

• Last year we build on 3.4 stream thru the spring maintenance
• We may need to build occasionally on top of 3.6 stream
• No dependent components require 3.6 from IBM Test project or from Intel. Some of the IBM Platform dependents that Kathy knows of will need both 3.5 and 3.6.

Our current strawman is to leave 4.6.2 on 3.5. For right now, we will dropping the 3.5 based driver after quickly testing it on 3.6.

Kathy will find out more about consuming requirements and we will discuss more at next PMC

There is a defect related to Itanium builds #286643. It relates to packaging a 32bit version of a driver component. If Itanium support will continue, we might need to get a 64bit driver built. Currently Yi is removing JVMPI support.

• Chris does not believe that support of J2SE 1.4 based JVMs on Itanium is critical on the Intel side. Chris to verify.

Misc

Paul noted that it is 09/09/09

Itanium build systems for IBM from Intel

• We continued the discussion about ongoing issues with getting the agreements in place for these systems. The PMC continued the discussion around picking a "drop dead" date after which the PMC will decide that Itanium support needs to be removed.
• Right now, the PMC is pencilling in the end of november penciled in for the "must arrive at IBM" deadline.
• Suggest that we should subtract about two weeks for the final date to get agreements finalized.
• AR to Kathy to validate whether IBM has any internal requirements for Itanium support that would cause issues if Itanium support was withdrawn starting in 4.6.2