Platform/2009-Q4-Goals

Performance: specifically, responsiveness.
- Create an automated responsiveness/interactivity benchmark.
- Improve Gecko/Firefox's score on that benchmark substantially (at least 25%).
- Support and/or participate in the startup time work being done by the Firefox group.
- Participate in the Electrolysis project by adding support for windowless plugins on Windows.

Improve the Web platform by contributing to tests and specifications
- Publish CSS3 Selectors as CR [fantasai]
- Ongoing development of reftest infrastructure for the W3C [fantasai, jwatt]
Improve the Web platform by implementing key features
- Land Harfbuzz prototype integration in tree (allows for richer typographic control, also lays groundwork for performance improvements) [jfkthame]
- Implement calc() (Web platform) [dbaron]
- Support CSS transitions/SVG animation on nearly all CSS values (browser parity) [dbaron, dholbert]
- Enable SVG Animation with shippable feature set (near-complete CSS property coverage, near-complete SVG attribute coverage, but not necessarily syncbase timing) (Acid3, browser parity) [dholbert, birtles, jwatt]
Improve perceived and measured performance
- CSS scanner performance analysis and improvements (performance) [zwol]
- Implement some kind of lazy frame construction [?] (still working out a plan)
- Compositor phase 2 --- paint and animation control (performance, also add JS animation API to the platform) [roc]
Improve core code architecture to improve stability/performance and reduce cost of future enhancements
- Simplify frame destruction [fantasai]
General product improvement
- fix scroll position restoration and maintenance (bugs 43114 / 103279) [dbaron]

Video on ARM - 15fps on N900 full-screen?
'buffered' and 'seekable' TimeRanges [cpearce]
Remove dependency on libfishsound, liboggz, and liboggplay [doublec]
Experimental implementation of accelerated full-screen video using cairo-gl [kinetik]

Resolve Mac OS X Java situation for Firefox 3.6.
Fix any major regressions in 64-bit Mac OS X builds.
Complete initial performance analysis of 64-bit Mac OS X builds. Should include comparison with 32-bit builds on 10.6 for major benchmarks.
Fix all known regresssions in Cocoa NPAPI, make it ready to ship.

XBL2
- Implement remaining binding mechanisms (rollover from last quarter)
- Implement support for mutations of childlists and the anonymous tree
- Implement xbl:attr and xbl:text
- Land document-audit patch on trunk
Move the web forward by implementing more of HTML5
- Get our HTML5 history API landed
- Finish the off main thread parsing work
- Speculative loading with the HTML5 parser
- Evaluate what parts, if any, of HTML5 to implement next.
Electrolysis
- JPW (cross process JS wrappers)
- HTTP over IPC, details TBD
Improve performance
- Staying on trace through same origin XOWs
- Remove the XBL1 CSS based binding mechanism for non-displayed DOM nodes
- Use our (almost) lock-free data structures for our main event queue
- Optimize DOM child storage for performance
- Minimize the number of XPCOM tearoffs we create from JS
General improvments
- Work with the security team to get the ball rolling on a enablePrivilege() replacement
- Improve our popup blocker to make users experience on the web friendlier.

Event handling performance implementation (for DHTML)
- Code reorganization
- Implementation of throttling and/or delayed computation and/or smarter caching
- Crashers
Fix keyboard support for HTML5 contentEditable
- Add accessibility support for HTML controls inside of a contentEditable
Polish table/grid/treegrid accessibility
- issues with some existing complex DHTML tables (e.g. dojo)
- vendor reported issues

Shippable multi-process Flash, windowed and windowless on Windows (and Linux)
Complete necko API revisions necessary to support multi-process networking
Usable prototype of Fennec running multi-process on Linux
Usable prototype of multi-process Flash on Mac OSX 10.6 (has external dependencies)
Completed gecko API changes necessary for multi-process session history and link targeting