Skip to main content
Jump to: navigation, search


COSMOS Community Call Minutes - February 28, 2008

Attendees: Jagmit, Martin, Domenica, Bill, Saurabh, Jason, Hubert, Paul, Jack, Jimmy, Srinivas, David, Don

QA Status:

  • Finished E2E testing with no blocking issues. QA team has found several non-blocking issues which have been posted to the dev mailing list. They will open bugzilla entries for each of those problems.
  • QA also confirmed that they can contain the Data Visualization manual tests during the i9 test pass.


  • Do we need to do any cleanup in the CVS directory structure? (e.g. The example MDRs are in 2 different directories. There are projects that are no longer used and code checked in that's not tested or not being maintained. There is data assembly code that we have not touched for a while and should probably be deprecated)
  • We need to do some investigation to determine what cleanup needs to be done. Perhaps we can talk about this on an architecture call. The smaller the list is when we release COSMOS 1.0, the better. We should try to make these changes before i11. Bill will cc Tania on bug.


  • Questions have been floating on the mailing list about testing with the IBM JVM.
  • There is no expectation for QA to cover this testing in i9. For i10, we need to figure out how to address this.
  • Several people have had problems getting the IBM JVM to work on non-IBM hardware (e.g. a Dell PC). This needs to be looked at quickly.
  • Srinivas will e-mail the details to the dev-list (Hardware, JVM version and error message). Srinivas should try installing IBM JVM on Linux, as this may only be a Windows issue.

Goals for COSMOS 1.0:

  • As we review the use cases for i10, we should consider our overall goals for COSMOS 1.0 since i10 will be the last iteration in which we deliver functional enhancements.
  • We should talk about Data Manager or Managed Data Repository - legacy repositories
  • Jimmy will post his thoughts on goals for i10
  • SDD - All of the SDD Tooling use cases have been moved over. The runtime use cases should be moved over next week (waiting on internal IBM doc). We want to refactor some of the use cases based on the doc.
  • For COSMOS 1.0 we will have SDD tooling and framework for buildtime generator. We are still trying to reconcile technolgoies on the runtime front (e.g. P2 - Equinox Eclipse project). We are trying to determine whether or not we can use the P2 runtime. SDD parsers will also be available for COSMOS 1.0. We need to transform SDD meta-data into the type of input we would expect. Also trying to figure out how to leverage some of the existing work in the MDRs.
  • There is currently no SDD entry in the development guide. We should consider adding this section.


  • Tania created a page to document team roles and responsibilities and included Sheldon's table of component owners within the subproject teams. Everyone should review this page and update as appropriate.
  • One role we should add is the "Early adopter" role. Jimmy will update the page with his thoughts.
  • We may also want to add roles for QA, RE and documentation.
  • We should think about the purpose of the sub-components we are defining within each subproject. We do not want to get so granular that we have to change the sub-components every time we get a new set of enhancements.
  • Since we are not a top level project, we do not have a PMC. So, we should be careful about using that terminology.

  • We should set up another walkthrough for the development guide - Tania will work with Rich to do this.


  • BOF - The expectation is that this is a group of interested people (would be happy to get 24 people from outside the project team). It is generally interactive. Our target would be potential adopters.
  • CA will be taking the presentation from a few months back and share their experiences with adopting COSMOS. Jimmy will review this with Mark.
  • David and Mark will be meeting this afternoon to review their short talk presentation

i9 Postmortem:

  • Weekly Integration builds were helpful
  • We had more functionality in i9, so that was one of the challenges
  • One of the snags was test coverage - e.g. QA trying IE when many of us don't use that
  • We could have moved to the Eclipse server in i9, but we did not want to destabalize the builds
  • We need to check with Saurabh on the status of the JUnits
  • In i9, we specified the linkage between the development and QA phases much better - have a more formal and better-tuned approach towards closing off the iterations
  • David sent an e-mail about holes in the testing - some of those will get hit in i10. Perhaps for i9 we can get this added to the performance testing phase. We should discuss this on the next architecture call.
  • We might want to have QA more involved in writing test cases. QA is planning to write both negative & positive testing

Back to the top