TPTP-PMC-20080423

Logistics

Attending: Paul, Oliver, Joanna, AlexA, AlexN, MikhailV

Ganymede I7

Oliver notes we should be in test pass 1 for I7. Whats up?

• Missing candidate driver
• TP 1 was supposed to be done by end of week
• Paul thinks that if get candidate driver today can get bulk of tests done (longer ones won't be done).
  • Oliver notes that the two passes are back to back.
  • Paul mentioned that this is normal in our plan

• Joanna suggests moratoriam on non-urgent patches to facilitate getting testing done
  • Approved
  • Goal -- get build by end of day today.

The build did complete last night

• There were some blockers opened last night: 228384
• There are a few criticals against platform
  • (duplicate host entries)
  • (reattach)
• 228029
• data pool test
• Paul noted that a consumer has come to him with another blocking defect that he is looking at
• Team should concentrate on above so that we can get test pass done.

AlexN notes that for monitoring he will be opening 1 more blocker (plugin stopped loading) and 1 bugzilla on a regression

AlexA talked for Trace/Profiling a bit about Harm's profiler use case. Use case is going pretty well. AlexA does have a few questions in to Harm about reproduce about reproducing other problems. One example is 214098.

AlexA talked about another issue: 212009

• Upgrading ARM agent to use new AC framework
• It is a critical issue initially assigned to Joe Toomey, now Richard
• Can't be in I7 because patch is needs some more work after code review.
  • Will need to move to I8
  • Joanna okay with I8 or if Richard is really productive putting in during I7 TP2
• Richard should continue to improve it
  • Joanna asks folks to please put concerns about the code into the bugzilla
    • Igor has comments as has Eugene already
    • Alex notes that several other defects are depending on 212009

Joanna noted a question from legal regarding JVMTI (Martini) files

• Question: Are files for documentation or for build?
• AlexA notes: They are for building documentation.
• Question: Where can the martini doc be found in build result?
  • It is primarily documentation that is for developers only available thru CVS but is not in GA release of all in one.
  • Joanna asks if it is available as part of JVMTI SDK available thru web site.
    • Alex notes that the resulting documentation should be in the SDK you get from the JVMTI SDK link
    • Alex notes that the Makefile, dep, dsp are only used for building and they do not insert content into the builds

SAP Discussion

Oliver and AlexA noted that Andreas had some questions/comments about this weeks AG minutes.

• In particular, SAP is interested in seeing extension points model to they can interface to existing heap profiler
  • They will consider implementing these extension points as their resources have time.
• They won't be able to take on existing heap profiling function

Oliver asks if we know enough technical details about SAP heap analyzer architecture that we can understand the integration complexity or is more needed?

• Do we understand the technology enough that can really discuss possible intersect points to TPTP?
  • We need to understand enough about the intersect points so that we can guesstimate the maintenance/enhancement overhead that is likely if we want to experiement with integration of heap profiler and heap dump analysis.
• Any update from Eugene about "taking SAP stuff for a spin"?
• AI to Joanna to make sure Eugene is ready next week to come talk to this at PMC.

Staffing changes discussion

Oliver spoke w/ Intel management chain earlier this week. Unfortunately, it was clear that resourcing changes are unlikely to change back.

Oliver wants to make sure we will finish Ganymede

• Chris notes that Intel is committed to fulfilling Ganymede deliveries

We spoke a bit about what will happen to project management structure when it is all done. We will probably need to roll new resources in under projects lead by IBM folks

• Harm and Joanna will need to close on the impact of this change on IBM's planning.
• Harm/Joanna are putting together a meeting at IBM with interested parties in the next few days to discuss.

Oliver noted that the June presentation to the board will totally change. A few weeks ago, the outline was:

• Try to grab community, POG, etc

Joanna states that will need to replace trace leadership

• Oliver raised possibility that we might move to more technical engineers leading subsets of work.
  • We didn't go into detail here. More discussion will follow in future meetings as we know more.

EPP packaging will probably be yanked

• There is general agreement that this is the right thing to do
• Oliver wanted to underscore the kudos/thanks to folks involved in exploring the possibility.
  • It was great work despite the fact that it won't go in!

PMC noted that how much profiler/AC maintenance will get done (e.g, bug backlog and responsiveness to new issues) will depend on how invested the incoming Intel resources are.

• There was a general discussion about how to encourage the motivation of the incoming resources.

Harm asked Intel for more detail about the resources that will be ramping into the project after Ganymede release and what point in time existing resources will go away. He would like to know what overlap is available for ramping. Intel could not comment about this yet.

Joanna asked how build work be handled.

• Team is currently documenting the processes and working thru how to facilitate transition.

MISC

Paul notes that we will need to create rampdown plan for Ganymede

• Joanna asks what needs to be in the plan
• Required by Eclipse (it was part of Europa last year)
• Paul will try to track down more detail from last year. Unfortunately, link is busted from last year.