Platform/2010-03-09

From MozillaWiki
Jump to: navigation, search

« previous week | index | next week »

Notices / Schedule

Firefox 3.0.19 / Firefox 3.5.9 / Firefox 3.6.2

  • code freeze at 11:59pm PT today
  • only two blocker patches left to land:
    • bug 504021 - mrbkap needs to land on 1.9.0.19
    • bug 546530 - roc needs to land on 1.9.0.19 and 1.9.1.9
  • several non-blocker patches have approval but have not yet landed:
  • QA to start verifications tomorrow morning

1.9.3 alpha.next

  • code in for Friday's nightlies
  • if initial crash data looks good, will give releng "go" Friday noon-ish
  • should have automatic updates if possible: want to build a base of alpha users

Blocker Report

Browser / Front End

  • the Firefox Projects page has been re-organized as have the pages for each project to make it easier to read
  • Extension Manager Redesign process; API & dev work outlined in wiki, met with QA to develop testing plan, design cases being added, including install from multiple locations & with various levels of security (tracking bug bug 550048)
  • The UX team has created a stream-of-consciousness style tumblr with design sketches, inspirations, and pointers to other bits of designery -- From The Bikeshed
  • OOPP, landed crashed-plugin UI for Alpha 2, working to improve some privacy issues now (bug 550293, bug 550303).

GFX Update

  • This was forgotten last week, but John Daggett finally published his font feature support work in the CSS3 Fonts spec.
  • Cairo update is getting close to landable.

Hardware Acceleration Update

  • Direct2D work is ongoing, with lots of bugs being found and fixed. The 1.9.3 alpha 3 we're going to release this week is aimed at a reasonable level of Direct2D support, albeit off by default.
    • One of the problems that we've found is that a lot of reftests assume that there's no such thing as subpixel positioning of text in the vertical axis. That is not so with DirectWrite support, and you should be aware of it.
  • Core Animation plugin support is coming along very well. We've got it working as well as is possible without OpenGL rendering support, and it'll only get better once OpenGL is supported via Layers.
    • We're working with the plugin-futures group to refine the Core Animation plugin spec to make it more implementable (and more performant).

Layout Update

  • Jonathan Kew fixed a large number of reftests to pass on Windows 7 (various font issues)
  • Ehsan imported BrowserScope's collection of HTML editing tests (code) and is orchestrating feedback on HTML editing in general (blog post)

Content Update

  • e10s event handling API (TabChildGlobal) starting to shape up (smaug)
  • structured clone for DOM workers and postMessage in review (bent)
  • Infallible operator new landed! (cjones)
  • enablePrivilege() is nearing EOL
  • File.url and nsIAtom UTF8->UTF16 changes landed (sicking)
  • AddRef/Release() on cycle collected objects now abort() if called off main thread (sicking)
  • HTML5 parser tp4 regression fixed (hsivonen)

JS

  • GC / Thread API changes http://bit.ly/a2mcB5
  • JaegerMonkey blogged and work continues
  • Still finding incremental SunSpider wins in the TM tree

Electrolysis

Startup Performance


Estimated win Bug # Summary Owner Status Notes
~10% "Dirty Profile" Startup Reduction Constant startup relative to profile size Shawn Wilsher
significant bug 558200 Extension profiler platform api Steve Fink There is an easy way to provide feedback to extension developers to help them figure out why their extension is misbehaving. We need to implement that ASAP, so typical firefox installs(with lots of extensions) can be fast


Security

Breakpad

  • At the platform workweek meeting, we agreed that we wanted to get our new work into M-C as soon as possible so that we could shake out problems in the nightlies.
    • We're currently using unmodified upstream sources, but the new work includes stuff that hasn't yet been reviewed by Google, so this means that M-C will contain the upstream Breakpad sources with local patches applied. I'll set up a shared Mercurial repo containing the patches we've taken relative to the public sources.
    • I'm finishing up the Mac dumper, which should be done before we bother with this.
  • Linux ARM: All work is done; we will need to take a local patch for this.
  • Linux x86_64: All work is done; we will need to take a local patch for this.
  • Mac x86_64: Not done. Ted has taken bug 548025 ("make Breakpad client libs work on 64-bit OS X"), and made it block bug 519060 ("[Tracking bug] officially support Mac OSX 10.6 64-bit builds").
  • Linux -fomit-frame-pointer: All work done; needs the same local patch as Linux ARM.
  • Mac -fomit-frame-pointer: Symbol dumper work needed bug 517832; this is my top priority at the moment.

Tree Management

  • Now doing Talos tests on Snow Leopard!
  • Downtime on Thursday bug 551205. 9-11am EST
    • Starting up Addons branch
    • All branches will be closed for the duration
  • 180 of 200 minis now online
  • TBPL now has linux64, will soon have 10.664bit
  • 10.6 64bit build being evaluated by josh bug 545539

Roundtable

  • Need help on memory profiling, Firefox 3.0-style. Stuart has volunteered to mentor someone on the mechanisms he used on 3.0. Need a volunteer to do the research. (damons)