TPTP-PMC-20080507

Logistics

Attending: Oliver, Harm, Paul, AlexA, AlexN, Chris

• Not Attending: Joanna and her representatives were not present

Any objections to previous minutes?

• No

Ganymede

I7 Finalization

• Currently there is no release candidate and we have not yet dropped (as of yesterday)
  • Harm notes that we had a candidate yesterday morning
  • We did not drop because of a defect (security fix needed)
  • Team believed they would be able to drop tommorrow
  • Harm suggested dropping yesterday's build as placeholder and then drop again.

What needs to be decided about I7 in this call?

• Harm's drop proposal
• Joel has May 6 driver ready to drop (this is the one from yesterday)
• Instructed to go ahead and drop it.

Oliver notes it seems a bit sloppy to be making these decisions a few days AFTER drop was due instead of a few days before.

• We should be able to do better.

Paul asks if closing I7 open us up for I8 checkins (this is supposed to be the development week for I8)

• Harm notes that the I7 test pass is not really complete
  • Looking at test pass report, a number of suites have not been run
  • Either test report is wrong or tests we intended to run have not been run.
  • Harm suggests that I7 test pass problems be fixed before letting people move on I8
    • Paul notes that this might mean that no development for I8
    • Harm is okay with this. He just wants to see testing finished (even if tests fail).

Test Pass Discussion

Leads have gotten AI over the last few iterations to ensure test reports to reflect tests that they intend to run

• If those tests are not being run as a project, we need to understand why
  • Tests need to be run
  • And/or leads need to fix the reports to reflect intent

Harm reiterated his position that projects that are ahead should help with test finalization of projects that are behind

• Paul reiterated his mixed feelings about this. Given some of tests involve specialization and deeper understanding it can be hard to fully do this in practice.
• Paul notes that his guys (guy?) can help a bit on testing

Query: How much time would it take to finish test pass?

• In general, all remaining is in trace and platform.
  • There is one strange one in suite in test that seems to be redundant or just in error or glitch.
  • AlexA notes that it will take about 2 days to finish Trace tests
    • Would mean that Richard can't do I8 development work.
  • Leads need to decide what is the priority, are the tests important enough to run or not.
    • If not, they should not be in the reports.

It was noted that finishing the I7 pass is increasing the risk of I8 defects (i.e., reduces number of defects that will be fixed in 4.5).

AlexN notes that can remove tests that won't have time to do from test plan

• Harm notes that these tests were there for a reason and that the leads need to be a bit careful about doing this just to get a clean test report.
• In some cases it may be the right thing to do (e.g., if tests are believed to not be relevant)
• but other times it is not the right thing to do
• Harm raised Trace testing as an example
  • UML sequence diagram views ~10 tests
  • These tests have not been run in any iteration in this release
  • We know that there are issues in that area (defects are happening)
  • What should happen?
    • Remove test from report?
    • Run the tests?
    • As/Is the component (sequence viewer)?
    • ...

Oliver suggests that we summarize the test status more regularly/fully in PMC.

• Till now, Oliver has simply asking leads if they are on track and not examining the detailed reports himself.
• Harm/Paul test reports refresh 2x a day morning and afternoon
• Oliver asks if test report is high level enough to be understood by mgt.
• Oliver suggested that someone needs to nag leads in between meetings to finish their test pass.
  • Paul suggests that we add a few more "penalties" of some sort for not finishing passes in time.

We seem to spend a lot of time focusing on fixing regressions and then get more regressions from our fixes.

• Better attention to test could help this.

We noted that when we lost resources last year that the time it takes to do a test pass went up (and could no longer be overlapped with development)

• Leads tend to be hesitant to really drop development to concentrate on test (they used to be able to overlap)
• In some cases we have estimated it takes 1.5 weeks to do a test pass but only allocate 1 week for it on the schedule.

A lot of defect fixes going in are without related tests being checked in

• Our balance is somewhat off between defect fixing and testing
• We are not as restrictive during test passes regarding putting features in as perhaps we should be
  • If leads think that something is putting a test pass at risk of not being done, they should explicitly involve PMC.

Defects falling out of 4.5

All leads had AI to create list of features that will slip out of 4.5

• AI was due by AG call May 5...
  • Not done (kudos to Paul for being closest)
• AI deferred to PMC call May 7
  • Not done
• AI deferred to end of week May 9...
  • AI TO LEADS TO FINISH THIS WEEK

Have all projects listed features that will be slipped out of 4.5?

• Test: list is up to date and lists features that Test cannot complete and want to defer.
• Trace: 1 is listed (paul put it in based on AG discussion).
  • AlexA was asked about more.
  • BTM support w/ new profiler interface will probably need to be on this list. (especially since Richard will need to run I7 tests instead)
  • Paul says that it is safe to assume that BTM stuff will NOT be done 212009 xxx821

Oliver asks when will list be completed about what defects will be deferred.

• Each lead needs to identify which issues need to be deferred that are currently targeted for 4.5
• When will AlexA have this list for Trace?
  • Hopes to have list tommorrow (May 8)
• When will AlexN have list for Monitoring
  • AlexN has been doing list the other way around (i.e., what CAN be done rather than what CANT)
    • Has 1 that is top priority [1], don't really care about remaining ones.
    • Has list of 4 defects that really want to see fixed in platform but needs to focus on one at a time
    • 2 other enablement defects need to be looked at.
  • When will AlexN have list updated?
    • Will update today. (May 7)
• Joanna needs to step in and update list for platform.

Harm notes that there are 100-200 P1 major/blocker/critical (targeting 4.5) across TPTP

• Need to be classified.

Chris picked up on a comment from Paul. We are approaching a time that this listing is irrelevent.

• I8 development ends this Friday (May 9) and pretty much everything will be deferred
  • With a few things priority pulled in on demand
  • As of now, it would be a shorter list to state what will be done rather than what will slip

Harm suggests that we need to have either list to HAVE or list to DEFER

• Inverse list can be easily generated.

I8 schedule

AlexN was looking at schedule and notes that development ends Friday

• Harm notes that we pretty much lost the week of development for I8 because I7 was not done.
• AlexN said that Monitoring has a list of defects with patches that he is ready to patch in
• Paul said that Test also has a list of defects with patches that he is ready to patch in

If we start testing next week, how will they have time to check in defects

• this is why we need to be careful choosing how many things should be in list.

Post 4.5 Discussion

Oliver summarizes:

• The best guess about resourcing from IBM at this point is "flat".
• There will be only a few folks from Intel and they will be new.
• it may be appropriate to just do bugfix for a while
  • Not confusing ourselves or being sidetracked with new enhancement effort

Since overall group is smaller, Oliver starting to think about organizing leadership into

• 1 management lead (possibly growing from platform lead Joanna)
• 1 technical lead (possibly growing from architecture lead Paul)

Oliver working on June presentation for board