Platform/2012-11-20

From MozillaWiki
Jump to: navigation, search


« previous week | index | next week »

Platform Meeting Details

  • Tuesday 2012-11-20 - 11:00 am Pacific
  • Dial-in: Audio-only conference# 95312
    • People with Mozilla phones or softphones please dial x4000 Conf# 95312
    • US/Toll-free: +1 800 707 2533, (pin 4000) Conf# 95312
    • US/California/Mountain View: +1 650 903 0800, x4000 Conf# 95312
    • US/California/San Francisco: +1 415 762 5700, x4000 Conf# 95312
    • US/Oregon/Portland: +1 971 544 8000, x4000 Conf# 95312
    • CA/British Columbia/Vancouver: +1 778 785 1540, x4000 Conf# 95312
    • CA/Ontario/Toronto: +1 416 848 3114, x4000 Conf# 95312
    • UK/London: +44 (0)207 855 3000, x4000 Conf# 95312
    • FR/Paris: +33 1 84 88 37 37, x4000 Conf# 95312
    • Gmail Chat (requires Flash and the Google Talk plugin): paste +1 650 903 0800 into the Gmail Chat box that doesn't look like it accepts phone numbers
    • SkypeOut is free if you use the 800 number
  • Warp Core Vidyo Room / SFO-Boardroom
  • join irc.mozilla.org #planning for back channel

Notices/Schedule

17 bugs
67 bugs
Unresolved Aurora Trackers (non-security, not tracked for Beta) Unresolved Beta Trackers (non-security)
  • 17.0 release is live (about to have update sign off & press push)
  • 17.0esr is live
  • 10.0.11esr push request sent, will be live within the hour
  • 18.0b1 going to build today, shipping on Monday 11/26
  • Aurora updates throttled, unthrottling on Monday 11/26

Products/Projects

Firefox OS

QA

  • Accomplishments:
    • Imported TEF testsuites into moztrap (calculator, ftu, utility tray)
    • Testcase count at 2137, about 65% completed to goal
    • Added more smoketests to the list, putting the count up to 40
    • Moztrap mapping of usecases is mostly complete
    • Pushed a fota update to stable dogfood users for the first time
    • Cleaned up the qawanted bugs in gaia this week
    • Payments (mozPay + trusted UI) has all MozTrap test cases finalized and signed off
    • Identity (DOM id) has some MozTrap test case work going on
    • Some mozPay + trusted UI testing has been executed - bugs filed and fixed
    • Started initial discussions to scope implementation and testing work for preloaded apps
    • Moztrap 1.2.5 has been pushed to production
    • Gaia UI automation work continues, with much improvement in framework
    • QA and A-Team continue to work through Gaia automation reliability + app-launching/switching issues
    • API Permissions Test Plan started
    • Permissions tests for this plan are being submitted in Bugzilla
  • Decisions:
    • Releng is doing a trial run of unagi builds via tinderbox. awaiting testing
    • Chatting with QAnalyst to ramp up more contract testers for b2g
    • Identity blocker list clarified and finalized
  • Issues/Risks:
    • 181 open blocking-basecamp+ Gaia bugs (Boot2Gecko product, all Gaia components)
    • 53 open blocking-basecamp? Gaia bugs
    • 121 open blocking-basecamp+ Core bugs (Core:All + Boot2Gecko:General)
    • 21 open blocking-basecamp? Core bugs
      • Note that these queries have changed to include all Core b-b bugs, so numbers may have jumped from last week
    • Still high reports of regressed nightly and stable builds
    • Card view is regressing a lot
    • Smoketests still not at 100% pass. lots of partial failures requiring reboots and crashes to get tests completed.
    • Performance issues around app switching, and app startup time
    • Packaged apps is broken on trunk
    • Though work on Gaia automation reliability improvements is underway (see under Accomplishments), it still slows current development

Firefox Desktop

  • Social: 17 is out the door! Next focus: multi-provider support, some polish on trunk
  • AppCoast is killing it on per-window private browsing bugs - refactoring many tests
  • Snappy bugs in action: Paolo has a plan to improve downloads incrementally and get rid of a lot of jank, and Neil is working on form history refactoring
  • Downloads panel: bug fixes from mconley just keep on coming, we're going to leave it on Aurora to get additional testing
  • Firefox Desktop weekly meeting: Firefox/Meeting

Firefox Mobile

\o/ We went GA with ARMv6 support in Fx17 (800MHz/512MB RAM)
  • Same blocklist setup as we use for Fx17 Beta.
  • ARMv6 min req's is: 512MB RAM and 800MHz CPU
  • Goal to Reduce minimum system requirements: Fx18 min CPU drops to 600MHz; Fx19 min RAM drops to 384MB; Fx20 QVGA support
Video Playback
  • Fx17 has support for system codes (i.e. h.264) for ICS and JB with a whitelist/blacklist
  • Fx18 has the same basic support
  • Patches in the works for (demonstrated at Layout/Graphics/Media work week) for HC and GB
    • unexpected bonus: it works on ARMv6
Performance and Responsiveness Improvements
  • Project 256MB bug 792131
  • Landed in Fx19 bug 812594 - PushLocalFrame in AutoLocalJNIFrame::Push taking more than 100ms when called from AndroidGeckoLayerClient::ProgressiveUpdateCallback on a Samsung Galaxy Q
Recent Landings in Fx19
Stability Wins
  • bug 810248 - java.lang.UnsupportedOperationException: at org.mozilla.gecko.GeckoEditable.length(GeckoEditable.java)
  • bug 811144 - java.lang.NullPointerException at org.mozilla.gecko.db.BrowserDB.expireHistory(BrowserDB.java:145)

Metro [Weekly]

Stability Report [Weekly]

Major Discussion Points

Socorro

  • In preparation of bsmedberg's hang reporting rework (the product side is in 18 and coming to beta this week), Socorro is now not recording the browser sides of hang reports since Thursday. Therefore they are fading out of the topcrash reports and we need to adjust the criteria for setting the topcrash keyword.

Desktop

  • No big issues making it across the channels this week
  • Nightly crash rate still somewhat high, but all top issues are fixed

Regressions/Top Issues

  • Beta 17
  • Aurora 18
    • bug 812985 crash in _cairo_d2d_create_brush_for_pattern. Core|Graphics regression that also affects trunk
  • Fx 16
    • 16.0.2 looks pretty good - the explosive report today doesn't raise any alarms.
  • Trunk 19
    • bug 813619 - spike in 64 bit builds
    • bug 813489 - new Networking regression
    • bug 813666 - crash in xul.dll@0x1726b80 | NoteGCThingJSChildren
    • bug 812203 - spiked on the 15th, but is now fixed
    • bug 813024 - Mac crash which is referenced as top crash by Scoobidiver, but has a lot of dupes - likely the same person crashing
    • bug 812636 - Web RTC crash which might be a dupe
    • The not yet tracked topcrashers are mostly Win64, which isn't tier-1 and we don't have on Aurora anyhow.
  • 'Flash 11.5 crashes'
    • bug 810797 and bug 807714 continue to rise as they're new regressions with 11.5 and this release is being picked up more and more.
  • bug 813205 crash in F21225463

Mobile

  • Crash rates for beta are up in 17.0b7 compared to other betas before, not because of new crashes but because crash reporting on Android 4.2 works again. Level in the same ballpark now as 16 had on beta.
    • Note: some of the crashing seems to be related to flash crashing on 4.2
  • We have some JB-only crashes, now happening mostly with 4.2 and Nexus 7, that we'll need to push for in the 18 beta timeframe.
  • bug 775142 is the topcrash on 18 and 19, apparently we are confusing localizers and crashing with their strings.
    • Steps to reproduce are listed in the bugs.

B2G

  • We need more testing of the crash reporting flow.
  • Biggest blocker atm is bug 811341 - we need to send reports when we're already online! (just landed on all channels)
  • We need testing around bug 811778 to make sure we follow the spec correctly.
    • We don't have crashmenow for B2G so it makes it a bit more difficult
      • There is a way to kill in ADB - that should generate a crash.

Snappy [Weekly]

  • Tab Strip
    • Results from the gfx/layout work week: tab switching improved by over 2x for both Facebook and GMail. (See changes in content below and Taras' writeup in his blog post.)
  • Content
    • Jeff Muizelaar sped up checkbox draw time by simplifiying border radius drawing (bug 809603).
    • Matt Woodrow sped up GMail by tweaking how we use layers (bug 811927, bug 811570, bug 811831).
    • Bill McCloskey fixed transplant GCs, which are about 75% of the worst GC pauses (bug 803376).
    • Bill also shaved a few ms per GC slice (bug 807535).
    • Olli Petay confirmed that we have reduced the median number of JS objects in the CC graph using Telemetry (bug 801719).

Games [1st Tuesday of Month]

Mobile Web Compat [2nd Tuesday of Month]

CritSmash [3rd Tuesday of Month]

This week we'll give the background of CritSmash. CritSmash is an engineering led collaborative activity involving engineering, security, and release teams with a focus on protecting users of our main products. It is an odd activity at Mozilla since it is one of the few places where our users are best served by us being carefully secretive (and leaving our unsung heroes unsung). We're working out final details on what data we feel is good to share during our Tuesday all-dev meeting.

The weekly CritSmash meeting really has two main functions:

  • ensure critical security bugs are dealt with in a timely fashion.
  • ensure fixes get into appropriate releases/branches.

We also have a weekly pre-CritSmash meeting where we rate security bugs (exploit analysis). If you have questions about either meeting please contact me (davidb) or your manager.

Memshrink [4th Tuesday of Month]

Engineering Metrics

Key Issues

Actions Last Week

  • jp to see about moztrap presentation in future weeks
  • JP to follow up on FF19 Gfx top crashes
  • Mossop to follow up to mailing list about super-review status, update doc and post link

Issues

  • Anonymous namespaces (jduell)
    • roc: "in Visual Studio 2010 (not sure about 2012), you can't set a breakpoint on an anonymous-namespace symbol *at all*"
    • Proposed language for C++ Style Guide:
      • "Anonymous namespaces: We generally prefer using 'static' instead of anonymous namespaces. This may change once there is better debugger support (on Windows especially) for placing breakpoints, etc on code in anonymous namespaces. You may still use anonymous namespaces for things that can't be hidden with 'static' (such as types, or certain objects that need to be passed to template functions)."
  • Future Issues - early discussion on dev-platform, monitoring for discussion if not resolved
    • Namespace style guidelines (khuey)
    • UA String Touch or Tablet
    • Not shipping prefixed APIs on the release channel

Actions

  • jduell to update style guide for anon name spaces and mail to list

Roundtable