From MozillaWiki
Jump to: navigation, search

Weekly Gaia Development Meeting

Announcements & Notices

  • We're all coming back from an awesome workweek in Oslo, lots of amazing work.
  • We branched 1.2, have we?



reporter: tony chung

The Run: The Results: The Breakdown: The Bugs (whiteboard "burirun1"): (all: 59) (new and reopened: 44) (resolved fixed, since testrun began: 8 )


Partner Welcome Pack finished Partner product training - work in progress First incidents report just received from Telefonica for Venezuela - analyzing it and will bring updates next week



Talking this week (9/17): lightsofapollo - periodic sync of email and notifications are in master - STARTTLS upgrading landed - integration test work - githubbot work started Notes: The Team:

   asuth, evanxd, ianliu, gaye, lightsofapollo, doliver, jrburke, gghosh, mcav, jhford,   cserran, pdol, tony (qa), william hsu (qa), robmac
   Backlog in Pivotal Tracker:

Media front end

- Talking this week (9/17): djf - Planning our next sprint on Thursday - Pivotal: - Team: djf, dhylands, dmarcos, johu, dkuo, jporter, mikeh, sikeda, cserran (epm), robmac (ux), patryk, marcia (qa), hlu (qa), skasetti (product) - Bugzilla Whiteboard Tag for tagging media related bugs for triage: [fxos:media]


Updated: Sep 17, 2013 Talking this week: bkelly

   TEAM: bkelly, geo, hub, huseby, jhylands, jld, kgrandon, mlee, wlach, Bruce Weng, Walter Chen
   Investigating 1.2 app launch time regressions:
   Meta bug here:
   First suspected general regression shown in datazilla
   Spent Oslo week investigating this
   Conclusion is that it is a testing artifact due to interaction between test delays and our pre-allocated process launch heuristic.
   Short term solution is to increase datazilla test delays between app launches.
   Long term solution is probably the Nuwa template process work.
   Unless partners are using similar automated delays, this is probably not effecting their results
   Automation team helped run eideticker tests to try to reproduce regressions seen by partners
   Many thanks to wlach for working over the weekend on this!
   Eideticker measures launch time to "first stable frame" using high speed camera
   Initial results do not show any consistent regression, although the results are a bit noisy.
   Continuing to investigate.
   Need to compare our test methodology with what the partners are doing
   Need to compare revisions between tests since our eideticker test run may have graphics fixes from Oslo that the partners had not received yet.
   Making good progress on power measurement automation with the battery harness implementation.
   Meta bug here:
   Native stack unwinding in SPS profiler landed last week.
   Add MOZ_PROFILING=1 to .userconfig when building b2g
   Triage: Component teams please tag performance bugs with perf keyword


Talking this week: Ben The team: blassey, bfrancis, dharvey, clord, cserran, naoki, hubert lu, krudnitski, francis, ian

   Event target fluffing has landed! Thanks to Dale.
   Tiled layers back end is trying to land
   Share URL should now work with email - thanks email team!
   Still a bug with Bluetooth web activities
   User selection of default search engine is going to be disabled by default and search suggestions is not going to be implemented for 1.2
   As usual you can see our backlog in Pivotal Tracker

Comms app

Talking this week: Anthony

   Nothing new since the Friday demo so…

System front end

Talking this week: gregor

   Bug tracking:

The team: cserran, gwagner, ladamski, reuben m, kyle m, michael h, aus, francis, jason, atsai, pdol

   Closed most user stories (2 open ones, 1 has failing test), Looking at possible regressions. (related to support and preloaded contacts runtime customization)
   Planning for 1.3 (Download Manager, persona signin, offline error handling, more powerfule notifications)
   Whiteboard tag - [systemsfe] - use this to get on the System FE team's radar


Talking this week: Mike Pennisi ( :jugglinmike)

   Welcome Yannick Assogba ( // irc:yannick_a) and Brendan McLoughlin ( // irc:bmac) to Gaia
   Landed Timer and Stopwatch
   A few changes left to be ironed out
   No duration since timer expired feature, need UX input
   World clock in 1.3?
   Investigating a performance regression in clock related to LazyLoad-ing
   Investigating a leo+ Alarm bug #908525
   Considering moving some code from clock/ to shared/
   Event emitters
   IndexedDB upgrade/downgrade abstraction
   Useful elsewhere in Gaia? 
   Working with David Flanagan to figure out our goals for 1.3
   Ideas welcome
   Still need devices for testing
   We have 2 non-developer build Alcatel One Touch Fire's, and 6 unagis
   We need 7 phones for 1.3 development


System platform


Media recording


Round Table

   Travis Build Results (and why we still need travis for fabrice)
   Tree Closures If you want to see how many unit and integration tests are passing/failing
   running tests on 1.2
   I've heard there will be mozilla-aurora builds for B2G, should we be running these in automation (UI/Performance)? What Gaia branch would these run against? We're currently running mozilla-centra/master and mozilla-b2g18/v1-train, would either of these need updating? We're limited on devices so we can't easily add more tests at this stage.
   Feedback from Tony Chung was that b2g18/v1-train should go away soon.  As we ramp up m-a/1.2 we can migrate the test resources over to there.
   Buris are very finicky and (for many engs) difficult to develop on. Are there better devices (like the nexus 4 perhaps) that we could test patches on?
   If you use ./, you'll break it. 
   We need a Windows only tool. That is painful for developers.
   We need ./ to not override Gonk
   OTA update for ZTE Open enables Fastboot - how do contributors take advantage of this to flash their own build?
   hamachi nightly builds not useful - need to disable gonk flashing in ./

Useful or Old Links


  • 7/22: James
  • 7/29: Ben
  • 8/6: Julien
  • 8/13: Anthony
  • 8/20: Dietrich
  • 8/27: James
  • 9/3: Ben
  • 9/9: (OSLO)
  • 9/16 Anthony