TPTP-PMC-20080903

Logistics

Attending: Joanna, Oliver, Chris

4.5.1 Discussion

Chris, Oliver and Joanna when she joined had a little discussion about how we ended up taking so much longer than we expected to finish the few defects that we planned for 4.5.1. A few points arose.

• The new IBM JDK kindof came in out of the blue and after the requirements had been selected. Team did not even know jdk was available
• Some queries from consumers arrived quite late (e.g., some questions around Windows 2008 support)
• In the future we should try to close on requirements lists more in advance
• Joanna also noted that resourcing shuffles at IBM and Intel did cause some transition effects.

Joanna mentioned that the last remaining defect is [1]. This appears to be a regression betweeen TPTP 4.4 and 4.5 that did not appear in testing because it only occurs sporatically yafter a number of iterations repeating the same test.

• A consumer is now able to run into it in a deterministic manner.
• May need to add some repetitive tests for sensitive areas.
• Asaf is helping out.
• Consuming product team is asking for daily status

For 244265, the IBM consumer ran 150 times with the IBM VM and got one failure

• Problem: Consumer did not start testing until I5 of TPTP 4.5. Otherwise we might have noticed this before 4.5 release.
• Earlier testing by consumers would be ideal by synchronizing schedules a bit better
• Problem: We would like to do more repetitive testing but we are resource constrained

There are a few other lighter requests that have come in. If 244265 takes a while, team may try to squeeze in a couple of others.

One recurring question releates to build configurations

• It seems that it is a bit sensitive to build configuration (i.e., some build configurations seem to mask the problem a bit more than others and make it more or less repeatable)
• Example: With latest China server builds, there are failures 10 out of 50 times.
• For 4.5.1, we will likely want to synchronize OSes and compiler versions between IBM and Intel.

Oliver asks if the 9/24 ship date is at risk

• Joanna says that it is still early and there is no immediate risk

Oliver noted that Harm has suggested that we do some code walkthrus by the experts in some particularly sensitive areas

• To get more eyes on
• To facilitate having multiple coverage in the event that an engineer is not available

Misc

Oliver finished a discussion nwith Andreas at SAP

• The Heap analyzer seems to be doing okay and incubating within TPTP does not seem to be desirable that this time.

Next week we will talk about Roadmap XML file owner

• Joanna should have a name for an editor at that time

We talked a bit about how discussions are going between Joanna and Intel team.

• Joanna is currently sending build related mails to Lei (replacing mails she previously sent to Kiryl)
• Joanna is assigning defects to Yunan
• Joanna did note that Chengrui had showed up on cc list for a few mails
  • Chris noted that Chengrui has now been assigned to project and that Joanna can assign profiler and AC defects to both engineers now.

Joanna noted that she is getting some winmail.dat files from Intel mails that she cannot read.

• It relates to Intel using Outlook/Exchange and IBM using Lotus.
• Joanna forwarded one to Chris who provided a link to a tool that will let her read the files.