TPTP-PMC-20090923

Logistics

Attending: Paul, Kathy, Eugene, Ernest, Oliver, Chris

Any comments or issues with last week's summary?

• No comments

4.5.2.1 M4

Team noted that the 4.5 branch is opening up for M4 checkins. Waiting for a few more fixes

• Hope for a good M4 build before China team goes on holiday
• Plan on a few more builds in October

4.6.1

Paul notes that we have all but closed on the release and wants to know if anyone has concerns about him doing bugzilla housecleaning.

• No opposition
• We Will be doing branching/tagging and opening 4.6.2 later today

4.6.2/4.6.3 planning

Kathy sent out updated schedule and we discussed it

• She added a few lines.
• We have a Nov 9 helios drop (4.6.3) so we need to pull in a quick smoke test a bit
  • 1 week earlier than previous version of spreadsheet
  • M3 test pass is smoke test
  • M4 test pass is more advanced testing

Paul comments that it looks like a good mix but asks about what the contingency is if we actually do have to start building helios separately because of an issue (like EMF last year.

• General discussion indicated that it would not impact schedule but would impact resources so it could effect amount of work we can get done in 4.6.2.

Paul asks if the test pass around Helios M4 is overlapped with 4.5.2 work or separately accounted from the development iteration.

• Oliver notess that we need to be quite conservative putting P1s into winter maintenance to accomodate possible interface issues to Helios. Should we note this conservatism in the schedule.
• Paul mentions that this is more a plan problem rather than a schedule problem and we just need to be rigorous on our planning

Kathy notes that at some point we will need to start building for Helios but are hoping to defer until toard the end of the year.

• Even if we do a test builds in Helios, we probably do not want to branch the codebase (split the stream) until that point to avoid having multimerge patches. (e.g., continue to drop builds based on Galileo)
• Kathy will add link to helios master schedule to her table

Oliver asks what would force us to start building on Helios.

• A problem like we had last year with EMF or other Eclipse low level API compatibility would cause it.
• Oliver asks that, if that scenario happened, what would prevent us from simply stopping dropping builds into Helios for a while then pick it back up later in the cycle.
• IBM will not need/use the builds until sometime in 2010.
• Paul unsure if we could avoid splitting the codebase in this case because we will probably have to start the engineering to resolve it sooner rrather than later.
• If we run into such a problem Oliver mentions that he will push for doing minimim splitting of our effort.

Oliver asks how many person-weeks of effort would it take to do the fork and create a new split Helios version

• Discussion suggested about 1 week to do the actual fork.
• The question came up about how agile/responsive we are when working in split streams
• The fact that many defects have to be "double merged" after the split is an immeasurable tax.

Oliver asks when IBM plans to pick up first Helios TPTP build and asks if we should we ask IBM to wait till Feb1 to look at the resulting builds.

• Kathy will look at the schedule and talk to some consumers to find out how this would be perceived.
• Oliver asks what is the ideal drop for us for us to start fully doing Helios based work
  • Kathy thinks M5 (~Feb2) is the best intercept for us.

Kathy saw request for Helios project plan (XML formatted mergable format). She saw the Galileo one and wants confirmation that we want something similar (since we are again a maintenance release)

• Oliver responds that yes we want something like last year; not too much effort should be put in.
• Team noted that last year we had to push API freeze later than we originally planned.
  • Do we want to go ahead and do the freeze later.
  • General guidelines are M6 timeframe unless individual project decides to slide it a bit.
  • Kathy thinks the best intercept for us might be M7.
  • Because we are a +2 project we have this flexibility because we do not have downstream extenders

Misc

We discussed the progress on Itanium build system transfer

• Chris will forward email address to Oliver so he can send message to Intel folks to be attached as discussion proceeds.

Kathy noted that we still need to close on the best time to start nightly builds to ensure they finish and can be checked before Intel China leaves for the day.

• Chris to start thread today.
  • NOTE FROM AFTER PMC: Chris had 1:1 discussion with Yunan who will send out discussion topic. A few hour shift will be helpful. Builds used to finish before China left for the day but they slowed down some time ago.

Kathy asked a bit about skill transfer from Intel noting that we have been doing casual discussions so far. When China team returns from holiday, we will want to do more of this.

• We probably want to do more systematic look at the documentation
• Chris suggests that we should get bugzillas filed on the areas where more docs wanted.
• Chris asks if we can also leverage existing weekly meeting that the team has set up.
  • Kathy notes that Intel China team has been documenting different pieces already and would like to find out if there is a list of additional items that Intel Beijing team is already planning
  • IBM team should also provide input regarding what else they would like to see documented.
  • NOTE FROM AFTER PMC: Chris had 1:1 discussion with Yunan who suggests that we coordinate in the regular meeting that has been set up. Chengrui has created most of the documents that he had on his near term roadmap.