TPTP-PMC-20080305

Logistics

Attending: Oliver, AlexA, AlexN, Paul, Chris, Joanna, MikhailV Harm was unable to attend.

Oliver congratulated everyone on a successful 4.4.1 and Ganymede I5. He then suggested that we quickly jump into talking about I6.

4.5 I6/I7 Discussion

Are there any other topics for discussion first?

• Joanna would like to talk about I7
  • Joanna wants to know if we could do something like we did for I5 overlap with 4.4.1 (with reducing test passes) to have more time for development.
  • Oliver has always thought that test was a bit heavy. Oliver would be fine with this.
  • Paul suggests that we have 3 weeks of testing for each test pass.
    • Reduce each to 2 weeks for I6 and I7.
    • Then leave 3 weeks for I8 since its all about test/release.
  • We still want to have a full test pass and a smoke test pass.
  • Do any project leads have concerns about containing testing in I6/I7 with 2 weeks each?

Oliver asks why we think that we can get the same amount of testing done in less time now.

• At the very least automated tests are already done for 1.4, 1.5, 1.6
  • Oliver notes that it would be great if we could demonstrate that automated BVTs are reaping this benefit.
• Joanna noted that Eugene and students are now on board and enabled to run tests
  • So platform tests can now be done in a single week.

Consumers at IBM are planning to pick up 4.5 sometime in March to start testing. We may start to see some filed defects from this additional testing.

• Oliver thinks that for the future we should talk to consumers and use their plans in order to gauge test/development overheads for the different iterations.

4.5 I6 and I7 Dates

April 4 is the foundation PII freeze date. That date is a constant.

• Move end of development for I6 one week later
• Make test pass 1 and test pass 2 each 1 week in duration
  • Changing from previous (1.5 weeks for test pass 1, 0.5 weeks for fixing criticals, 1 week for test pass 2).
  • Everyone [should have] defined what their test sets are for I6.
    • Will this reduction in time impact success of test pass?
    • Nobody mentions a concern.
• AI Paul to update detailed schedule on site.

Paul has a question about pushing out I6

• Does this mean that translation stuff can be pushed out as well.
• AlexN says that these cannot be pushed out more than a few days
  • Final Deadline for translations is the 21st (to be uploaded into database).
• AlexN is okay with minor slip here as long as people are okay for the official deadline.
  • AI to leads to have them in to Alex by no later than the 19th to give him the opportunity to process and upload them.

Automated Gui test limitations

Alex notes that AGR tests need to still be run.

• AGR are NOT in BVT today.
• Team is actively working on getting AGR going with BVT
• Hope to get it done for I6 test pass
• Paul does not see a difference between manual execution of AGR and BVT execution of BVT (for XP platform)
  • Alex notes that sometimes AGR can fail because of platform specificic behavior (slightly different speeds, etc).
  • Paul sees this as a good thing for BVT because it will be a forcing function to the test developers to go and fix their tests to make them more reliable.
  • Oliver talks a bit about the cost/value of making your tests really good.

I6 Status Update

Oliver asking each project lead about I6

Test

• Test cases are set up
• Team is somewhat on track
• Have more defects in schedule than have resources
  • Are gathering list to move out of I6
  • Only 2 weeks of development time in I7
  • Will soon have list of defects to move totally out of 4.5
    • Paul believes this list will not involve consuming product depenedences
    • Deferred items do not have active CC lists for the defects
    • Things that would have been nice to have.
• Oliver notes that next week in schedule looks quite busy
  • Paul notes that with the extra week of development those items will spread out a bit more.
• IBM consuming products filed ~130 defects and about 95 are fixed already.
  • ~10 deferred to future

Oliver notes that if some bugs are being fixed filed by community that it could impact effort put in to fix defects filed by internal consumers at IBM

• How does Joanna make that call?
  • Joanna has weekly discussions/mail threads regarding defect backlog with consumers
• Team takes into account severity (e.g., if there is some low priority consuming product defect to tweak something)
• Oliver notes that this is actually a value for the consuming products in fixing community discovered defects because then the defects are fixed in advance of IBM consuming product customers running into them.
  • Joanna notes that most bugs are found in our test passes well in advance of consuming products even trying it out.
  • Oliver may want to stress this value of community testing with consuming teams in the future.
  • Oliver mentioned previously on PC meeting
    • Some ISVs saying bad stuff about lack of value from community
    • Oliver noting that community as a testing vehicle has great value.

Monitoring project

• Team is making progress on the defects
• Many of them are targeted for next week
• Two have slipped from first week of I6
  • These Require build changes (waiting in a queue to be processed)
• Remaining slipped defects will be done this week.
• Team has 3 defects related to updating automated gui tests
  • Failed starting in I5 because of some refactoring
  • Needs fixing before I6 test pass
• Concentrating on P1 defects and on defects that IBM consumers are stressing
  • Quite possible won't get everything done but right now it is looking fairly containable.
  • May have to eventually make the call to drop the lowest priority ones
  • Hopes to have clarification on lists from IBM consuming products in next few days to help with priorities)
    • Oliver asks if Joanna is hooked into this? She is.

Trace

• Oliver sees some bugs being fixed and is excited about it since its been a while since work for Trace appeared in the schedule.
• 6 issues in bugzilla to close in this iteration
  • Two are critical
• There is also a documentation issue. Need to update 2 tutorials about BTM.
• AlexA notes that there is still no response from Joe Toomey
  • Would like to have direct communication w/ Joe but no response yet.
  • Alex only has indirect information from Richard stating that he is working alone and no deadlines are scheduled
    • It is hard for AlexA to predict success
    • These defects do block another defect as well

Oliver asks IBM guys whats up with Joe

• Joe is about 10% on this effort
• Joe has been working on a product release
• Joe has not scoped it the effort himself
• Tivoli guys are giving Richard the AR to provide details to Joe who will then size it.
  • Original response from team was that they could not do it.
    • Joanna pushed to get them to size it first and give a realistic estimate.
  • Oliver asks how important is this? Should he grovel/push for this?
    • AlexA notes that IBM should find BTM updates for JVMTI are important (required for Java6 to work). If IBM cares about BTM working on Java6, they should care about this.
    • Joanna believes it is quite important as well and will continue to work this internally.
  • Problem is that if it does get done, it probably won't get done until I7 because of Joe's product delivery
  • Some flexibility next week because Joe is finishing a product release this week.
  • Oliver asks if we can we delay another week
    • AlexA suggests that some discussion from Oliver may be good now because Joe has given no feedback yet.
  • Oliver asks how bad would community be hit without this feature
    • IBM would be most affected.
    • Intel has no requirements around BTM
    • For now it is up to IBM to decide whether or not to fix it. Joanna will follow up.

Platform

Joanna noted that the team had an IPv6 meeting yesterday

• Jonathan made a proposal about supporting it in stages
  • Proposal reviewed by all involved parties.
  • Has a list of defects involved
• May require some I6 defects that slide into I7
  • will need to revisit that in a bit

IBM Ottowa contributed static analysis some time ago

• But they don't have time to work on it right now in community
  • IBM guys have internal codebase and external codebase and have been concentrating on internal codebase
• Their code doesn't build with latest CDT version
  • It caused some build failures last week making us roll back CDT version in use
• At very least team needs to get code updated to work with new CDT
• Joanna has asked Harm to talk to them about fulfilling their commitment to the OS community.

Any more I6 items?

• No.

Meeting at EclipseCon

We have a room the whole day

• Joanna cannot come
• Oliver asks if she can provide some dial in times
  • Joanna will provide this.
• Paul notes that he accidently sent the meeting invite to a wider distribution list than intended.

MISC

We had a quick POG update

• Oliver asks if I7 will be better for our use cases than I6?
• AlexA notes that team has been working on it and filing defects related to improving AC and UI experiences
  • Even if do not get significant improvement in profiler experience are already getting some benefits.
• Still discussing AC configuration improvements.
• Oliver asks if we will be proud of what we hope to deploy into WTP and if we are ready for this step.
  • Yes. It has some problems but is definately more useful than before. We need more users to get their feedback about future directions.

Oliver asks if AlexA believes we should be in WTP?

• There will not be more HC appearing for support. So if support questions come in it will be traded off against other maintenance activites.
• Oliver tries to explain why investing more heads in community support actually helps you if you are going to be releasing a product.
  • Regarding support overheads from his previous experiences... Tend to get lots of questions of people with pretty basic problems who are not reading documentation. Question about how to deal with this class of things
    • May need to update some documentation/tutorials.
    • Updated web content for profiler.
    • This type of thing may reduce "dumb" questions.
  • Mikhail Voronin mentions that some folk will still be lazy and not read documentation
    • Mikhail suggests that the "optimal" settings for a given release should be set by default and not require manual configuration
    • Mikhail raised an example of wanting to enable binary mode by default in 4.5

Oliver asks if we are planning to tune the defaults to be the "best" before WTP release

• Is this sized within the team?
• Team noted that the it is not a high overhead for setting good profiler defaults
  • Some items will still require customization to get good data (e.g., filters)
  • Chris notes that there may need to be some discussion around changing defaults
    • For example we had discussion already about binary formats and some members hesitated to change the default away from legacy mode.
    • Oliver would like to see us not getting too mired in legacy for optimial configuration
    • We need to close on this.

Question to Joanna about whether this (WTP) will cause large support burden that we can't afford.

• Joanna notes that we are already showing resource stretching from IPv6 and Java6 support (we are having issues getting them done)
• The thought of any additional overhead is a bit frightening.

Oliver asks whether some test stuff should be stressed in the WTP bundle as well.

• Paul thinks it should be.