CBI/May1 2012

Meeting

• May 1, 2012, 10am EST
• Eclipse Conference facility

Attendees

Thanh Ha, Andrew Ross, Carl Andersen, Paul Webster

Minutes

Agenda:

   Starting to look at WTP - Thanh & Andrew
   374438 script for generating pom.xml files - Thanh
   377410 compare test results from PDE & CBI platform builds - Paul & Thanh
   377190 compare IU's from PDE & CBI platform builds - Paul & Thanh
   Pushing CBI platform build changes upstream & decommissioning the CBI fork repos (aka. Bug 376038 - repository copies from CBI are very misleading)

Andrew: our approach for WTP will likely be similar to what we did for the platform... create a temporary fork to work on adding pom.xml files and parent poms

Paul: Jar binary comparison, John sent class in p2 tools jar, write up a class for given m6, ignore qualifiers and compare + do a report same files, method signatures, etc. Paul will post to a bug.

Paul: igor's tool may do the generation

Thanh: 377410 - test output results published to build.eclipse.org for M6 Paul: Need to compare against m6 test output, he won't have time to look at it for the next while Thanh: just look at summaries in browsers to stare and compare? directory diffs Andrew: diff the directory may be surprisingly helpful, you can mentally ignore timestamp diffs and spot real ones Paul to post link to M6 test output to help Thanh find it.

377190: Paul: shouldn't be a diff between IU's. once we get past, compare qualifiers. Not sure the qualifier comparison will be possible even once Igor's done the qualifier tickets unless we force the qualifiers to be the same.

376038: Once pom.xml files are pushed up, how often do we need to change them?

Thanh: When we rebased to M6, we didn't have to change many pom files... just fixed version for situations where they'd changed.

Paul: last commit will contain the pom files thus update all qualifiers once we push upstream. If the pom files go upstream prior to Juno SR0, then we'd avoid resetting all the qualifiers. <=

Andrew: Check Igor's signing/packing work, and bring to PMC & Architecture

Paul: an alternative would be to manually fix qualifiers for SR0, and have each component un-fix when they need to generate fixes after SR0

Andrew: Aggregator repos? Paul: I thought that was a CBI offering? Andrew: We'll make it so. Note: Emailed to cbi-dev, no response. This just makes sense so we'll recommend it from here on.