- Continue supporting Releng in Android:
- Complete Android Talos. Success is Talos running hands free by end of quarter
- Begin Android unit test rollout. Success would be mochitest browser-chrome running in staging by end of quarter? (Ideally though, if browser chrome works, then all unit tests should work because it's identical buildbot code except for the python command line to start the tests).
- War On Orange:
- Finish UI bugs (TBD: Which bugs block v1 of the UI? Those are the ones we should fix)
- Complete move to Elastic Search Database
- Drive the war on orange effort this Quarter. Success is: Working with the developers to have eliminated the top 5 oranges by end of quarter.
- Addons Testing Service - what's left to do here?
- Bugzilla Improvements:
- Complete splinter
- Complete/Evaluate (TBD: what is success) pulse extensions
- Complete/Plan (TBD: what is success) Bugzilla 4.0 upgrade on BMO
- E10S - move tests to use specialpowers support
- Decide on proper mechanism for special powers
- Transition tests to use specialpowers, success is transitioning all mochitest plain tests to use the new API or to be chrome-tests (we will work with developers to do this)
- Performance work:
- Release first version of E10S user responsiveness measurement test
- Fix ts_cold test
- Create an interface abstraction in the buildbot code for test infrastructure/help with mozharness (needs more specifics)
- What do we do here: Manifests for test harnesses - complete xpcshell and add support for one of the mochitest variants
If you're blocked from working on one of the above, take a crack at these...
- CrossWeave for mobile - investigating, any news?
- Crowd-source testing for Fennec, lower priority than android support and manifests
- Replacing Tscold?
- Release Mozmill 2.0 branch this quarter (still need to meet with team to determine what we'll release - a beta or a final)
- BugHunter - support existing system for firefox 4 release, move to new database backend
- Create a test automation system to test our own code
- Work with releng and metrics to help get rid of tinderbox
- Create a way to display test results that are not run on every checkin. This is a way to visualize a test run that is performed on a range of pushes - the set of pushes from the time of test start to the push just after the last test run. Examples of these tests-- valgrind tests, the mozmill update tests, cross weave, firebug etc. More Information.