Notices / Schedule
- shipped fix for CanSecWest p0wn20wn vulnerability on Thursday, April 1st. 8FD.
- release notes: 3.6.3
- so far, regression free!
- shipped as scheduled on Tuesday, March 30th.
- this is the last planned support release for Firefox 3.0 / Gecko 1.9
- release notes: 3.0.19 | 3.5.9
- only 3 open blockers at this time
- LegNeato has a proposed schedule for shipping Lorentz in Firefox 3.6.4:
- Thu, April 8 - Lorentz beta available for public download (see below)
- Mon, April 12 - Firefox 3.6.4 code complete w/OOPP backport landed
- Wed, April 14 - Firefox 3.6.4 beta builds complete
- Fri, April 16 - Firefox 3.6.4 beta offered as advertised major update to 3.6.x beta channel
- Tue, April 27 - Firefox 3.6.4 beta offered as minor update to remaining 3.6.x beta channel users
- Tue, May 4 - Firefox 3.6.4 shipped
- 5 open blockers
- schedule tbd
- builds now available
- plan on announcing them and asking people to download and test later this week
- today or tomorrow's nightly
Browser / Front End
- Cairo update landed & bounced last night.
- GFX Q2 Goals - please comment in dev.platform if you have questions, issues, requirements, or thoughts!
Hardware Acceleration Update
- Hardware-accelerated full-screen video using OpenGL (Windows only) is finally landed on m-c and working on most machines. Bas fixed some bugs last week to ensure we weren't hardware decelerating video.
- Jonathan Griffin has started finding bugs using the Grafxbot hardware acceleration testing extension. This is good news!
- Layout goals
- Mostly about layers integration
- Matt Woodrow has patches to add new cairo_retained_path_t API, implement the API to retain CGPath objects in the cairo-quartz backend, and retain paths in SVG
- Chris Pearce landed new nsOggDecoder. Tracking a few regressions but things are mostly good or better.
- David Baron landed fix for :visited privacy leak. Looks good. Need to track site breakage and consider whether we should have special handling for same-origin links.
- Jonathan Kew reports that his Harfbuzz text shaping backend appears to give the same performance on try-server as the current Windows backend. The former is doing kerning and ligatures on all text while the latter is not, so this is actually a significant win --- if Tinderbox agrees on landing!
- Chris Double landed new YUV code but it bounced due to build issues. Working on it.
- Robert O'Callahan's retained-layers work reached a milestone; it can now retain layer trees (but not layer contents) between paints and pass reftests on Mac.
- Daniel Holbert working on <animateMotion> to animate motion of SVG content along a path.
- Content team goals are in the wiki
- Peterv landed fast unwrapping.
- Bnewman implemented variadic C function support for js-ctypes
- mounir making good headway on HTML5 forms, meeting with web developers to know their needs, etc.
Platform-specific Support Update
- TM Merged last night, after a long break
- [josh] If you have Flash 10.1 or JavaPlugin2 installed on Mac OS X 10.6, trunk builds will run them out-of-process by default. There are some significant known bugs, but we think the exposure will help us resolve issues quickly.
- please file those bugs in Core::IPC and cc :BenWa
- you can get Flash 10.1 rc from Adobe Labs
- Posted a performance roundup last Friday.
- Vlad has a new patch up for the startup timeline (bug 480739), needs review from bsmedberg.
- I put up a patch on bug 557552, which adds the ability to compare performance data between two pushes to TBPL.
- View snapshot results for startup and all other tests.
- More details on the project page.
|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|
- Macintosh OS X symbol dumper code written. bug 517832
- Mac symbol files now include CFI, needed to enable -fomit-frame-pointer on Mac.
- Mac and Linux dumpers now share all but top-level and ELF/Mach-O code.
- Should correct 64-bit problems.
- 19 patches in Google's review queue.
- Ted will be taking paternity time off any day now!
- new talos dromaeo suite enabled
- now running unittests on 10.5, 10.6
- doing on mozilla-central now, will roll out to other branches while watching load
- will turn off old unittests-on-10.5-builders soon
- maemo5 gtk and maemo5 qt builds enabled in production last week
- 10.6 64bit builds enabled in production any...minute...now!