From MozillaWiki
Jump to: navigation, search

« previous week | index | next week »

Notices / Schedule

Firefox 3.0.16 / Firefox 3.5.6

  • builds made, on track for mid-December release

Firefox 3.0.17 / Firefox 3.5.7

  • proposed schedule online
  • tree is open!
  • please work on your blockers

Firefox 3.6 Beta

  • 500,000 active daily users (though only 50% are on the latest beta)
  • need to get to 650,000 to be able to map stability data onto the general population

Firefox 3.6 Release Candidate

  • there are 18 code blockers remaining
  • goal is to get to RC build this week, long poles appear to be
    • GFX kerning bug required by Fennec
    • JS new bugs reported from beta 3 rollout, need to wait for beta 4 data to come in

Blocker Report

See more Firefox 3.6 related blocker queries, or learn about the new status and blocker flags

Browser / Front End

(Progress reports every weekend on Planet Firefox)

  • see our active projects and get involved / propose others
  • Namoroka/mozilla-1.9.2 front end development:
  • mozilla-1.9.3 front end development:
    • Unfocused has tryserver builds up with tab match in awsomebar
    • David Dahl has a new Places query API
    • Mossop is updating the extension manager API and Boriss is looking at the UI
    • Faaborg and the Places team are starting into the UI changes they proposed and discussed earlier
    • Rob Strong is analysing several potential updater fixes
    • Limi is looking at the resource packages issue
    • Paul is updating the tab opening order heuristic
    • Marco is moving Places expiry I/O off the main thread
    • Drew is making asynchronous containers for Places operations
    • Horlander is driving forward the theme change work

GFX Update

  • 3 blockers.
    • The long pole continues to be Stuart's subpixel positioning/kerning bug (latest ETA: "later this week").
  • 0 noms.
  • Jim Mathies' out-of-process windowless plugins on Windows patch is undergoing review, as is Bas Schouten's Direct2D Cairo backend.
  • Bas has had quite a number of downloads of his Direct2D build of Minefield, providing a lot of great feedback and bug reports. Unfortunately some of the bug reports are of the sort we feared, driver- or hardware-dependent. We're hoping to get Direct2D into mozilla-central as soon as possible, preffed off, for wider-spread testing.
  • Bas also posted a great writeup on measuring performance of his Direct2D builds: how to do it, how not to do it, and when it's useful.
  • Chris Jones' Electrolysis shared memory support is undergoing review, for future use in Fennec+Electrolysis. Work for shared memory tilebrowser is waiting on this.
  • The WebGL standard goes to Khronos review this week. More public information will be forthcoming once the Khronos board approves it.
  • Topcrashes and blockers are consuming the rest of our time.

Layout Update

  • 0 noms
  • 2 new blockers
  • bug 527280 looks like it could be Flash video randomly corrupting browser memory ... very difficult to reproduce/analyze, could use some help trying to reproduce on Windows
  • SMIL syncbase timing patch submitted [birtles]

Content Update

  • 1.9.2:
    • 3 blockers, all crashers
    • 1 nomination, needs more information
    • bug 520639 has a likely fix, the rest are being investigated.
  • bnewman demoed JPW's talking across processes.
  • peterv fixed a cycle collector top crash for 3.6.
  • Jonas has been investigating crashkill bugs and evaluating what dll's to block etc.
  • hsivonen was able to land the fix for a HTML5 parser crash fix, and continues to work on the off main thread parser.
  • jduell has primitive HTTP working over IPC.

Platform-specific Support Update


  • 7 blockers left needing patches.

Startup Performance



  • Rob Strong put up a detailed post about his startup improvements to the update service, the effect on mobile, and points out some other front-end services that could benefit from the same treatment.
  • Joel's static Firefox (sans nspr and nss) is working (bug 525013), and is showing ~10% improvement in startup time on Mac. He's moved on to improving code locality (bug 531406). However, time-ordering of symbols doesn't seem to improve the sequentiality of the page-ins by much, so maybe more improvements there. More data on the bug.
  • Ben is making progress on the fastload replacement in bug 520309. Some fastload/invalidation discussion in this dev.platform thread.
  • Taras has patches up for service caching (bug 516085, needs review bsmedberg.
  • Super-fast-path-ing of Components.* needs some JS team help: (bug 512584).
  • bug 519445, for reducing Mac startup time spent in font system initialization, is ready to land when the tree re-opens.




  • Plugins: landing preffed-off in mozilla-central as soon as practical. Two potential blockers:
    • need mochitest-ipcplugins enabled on mozilla-central, bug 530522 (catlee)
    • Tp4 test intermittently hanging, possibly tab-related and doesn't need to block (smaug)
  • blockers for turning OOPP on by default, especially note
    • Talos process tracking, bug 529137... needs review by Alice and then landing (will require downtime)
    • bug 531860: .drawWindow with OOPP fails, needs owner
    • bug 519541: need to collect crash statistics from the plugin process; don't necessarily need the final UI, but stability statistics will be very important... ted do you have time?
    • Would like to get this turned on and an alpha/beta out by mid-December, though workweek activities may delay that.
  • Networking: jduell has channels working in a primitive way (not yet landed): beginning to ramp up other help (fred) to help hook up loadgroups and get fennec working. Honza posted initial patches for async redirect handling in necko: bug 513086

Tree Management

  • Help with making debug tests green on mac and windows bug 523385
  • 3rd try: this week will reset try repo.
  • Thanks to all those who traded in their old minis. We added a bunch more to Try Talos, and have enough left over to start work on Talos on 64-bit Linux.
  • 40 new n810s added, bringing our total to 80.


  • Firebug - 2 open blockers, one with a fix in place.
  • Blocklist reminder - channel all blocklist requests through the existing component. We prefer to block things at the plugin/addon level instead of the DLL level, so let's have everything go through the same channel for the time being.