My notes from OMR Architecture Meeting (please add any corrections or things I missed!) :
No additional input on PRs that we need or issues we need to resolve. We defaulted to picking the same commit that OpenJ9 has currently chosen for its 0.17 release (which has already been branched).
We discussed what we should do if OpenJ9 (whose release date is later than Oct 4) should have to fix something. We decided to "hotfix" the release with an additional commit (and different tag for e.g. 0.1.1?) so consumers can choose to take such fixes or not. Dan Heidinga from OpenJ9 thinks it unlikely that OpenJ9 will actually use this OMR release branch (at least this time around) due to the complexity it would introduce into release management, particularly given they have already split for their next release.
Everyone agreed we would be a source code release with the only binary being a Linux x86 JitBuilder.tgz package containing libjitbuilder.a, the include files, code samples, etc. Mark will be responsible for assembling the release package.
Daryl agreed to handle the release branch and tag creation, as well as creating the GItHub release information.
Mark will assemble the release notes, mostly by lifting from the release plan and probably borrowing structure from Eclipse OpenJ9 releases :) .
No additional testing will be performed as part of this release, but for future releases, Shelley Lambert mentioned some work she is hoping to get to in the near future that we may want to incorporate into our release testing regimen:
- GC performance tests to get back up and running
- restart code coverage testing
Follow on discussion for next architecture meeting is to start thinking about how we will continue the release train: when will it be, how will we decide content, etc. We did agree to start discussing it ahead of time rather than this release which is really just getting the last 3 years of work "out the door".
