Platform/2008-07-02

« previous week | index | next week »

Notices

• Propose a new meeting time of 2pm PDT rather than 11am PDT. This will allow NZ and Japan to dial in. They currently don't join (except for roc) as this is 6am and 3am for them. Objections? Thoughts?

GFX 1.9.1 Update

• GFX blocking 1.9.1+
  • vlad finishing up border work, need to find machine to reproduce perf problems
• GFX wanted 1.9.1+
  • SVG fonts
    • karlt: no progress due to distractions. This will benefit from the infrastructure being developed for downloadable fonts.
  • Do we have an update on bz's work to make <img src="foo.svg> possible?
• Video (roc/cdouble)
  • Core element patch uploaded, needs content review now, hopefully land soon
  • Some tests written, still working on them
  • Ogg Vorbis/Theora backend could get review soon after core element patch lands
  • Probably should try to land GStreamer and Quicktime backends soonish
  • DirectShow backend playing some movies, needs more work
• downloadable fonts (jdaggett)
  • unified font entry in this week hopefully (bug 437356)
  • starting on user font object (bug 441473)
  • zack: @font-face parsing is coded, but does not populate any data structure yet; am planning to make it build a standard CSS DOM structure; the user font set object should then populate from that; see bug 441469 for details
  • roc: this (and SVG fonts) seem most at risk; hard to see it working by end of July
  • jdaggett: SVG fonts is a much bigger task than downloadable fonts (parsing, external refs, shaping, drawing, etc.)
• pixman/cairo perf (vlad):
  • Will do new drop of cairo in next 2 weeks
  • Will have SSE2 opts and additional work from antoine/fred
  • 16.16 pixman coordinate space still an issue

Layout 1.9.1 Update

• 1.9.1 Layout Bugs
• Acid3: Roc or dbaron, status?
  • Nothing much happening, lots of patches waiting for review (dbaron, bz, jonas)
• SMIL still @ risk? (dholbert)
  • May not have a 100% complete implementation, but we should have a good subset implemented in time for feature freeze
  • Currently working on integrating with style system to animate CSS properties
• SVG fonts at risk?
  • karlt: yes.
• SVG CSS, roc:
  • Patches trickling into trunk, need review mats immediately and will need reviews from mats and rlongson and others for ongoing merging
  • SVG gradients and patterns as backgrounds for HTML elements --- working in branch
  • Starting work on feature to use an element as background for another element (covers -webkit-canvas, (combined with transforms) reflections, thumbnails and other use cases), should be quick as infrastructure is in place
• white-space:pre-line (roc)
  • needs review
• box-shadow (Ventnor)
  • Style system changes need review dbaron, rest ready to land
• CSS transforms - keith:
  • Parser can read -moz-transform property. Working on helper -moz-transform-origin property.
  • Basic rendering works (e.g. can draw all transforms). Need to work on computing overflow rectangles.
  • Working on mouse interaction and click detection.

Content 1.9.1 Update

• Content 1.9.1 Bugs
  • ~30 wanted bugs, needs triaging. Nothing critical for alpha1 on the list from first look.
  • Worker Threads, bent, status?:
    • Ben will work on getting worker threads code split into reviewable patches. Basic message passing, start/pause/resume/cancel, and (mostly) timers working so far. Once that's done he'll continue with what's left. Most recent WIP patch in bug 437152.
    • Spec about ready for general consideration. Arun to arrange meetings with Gears folks, others.
    • DOM worker thread XHR (minus the 'X') not yet started.
  • Tab reparenting (bz), at risk for 1.9.1? [I'm hoping to get some work done on this in the two weeks starting July 7. We'll see how well that works and whether it's enough. I think it might well be. --Bzbarsky]
• Cross-site XHR, at risk for 1.9.1? What's the current status on the spec and us internally agreeing on what to do?
  • Jonas in Redmond this week with Arun discussing how to proceed here.
• Critical plugin infrastructure bugs ready to land (reviews pending).

JS 1.9.1

• JS 1.9.1 Bugs
• JS Tracing:
  • Last week's status:
    • Tamarin nanojit integrated in SpiderMonkey
    • Recording traces for single paths through simple loops
    • Cooperating with Adobe folks on nanojit API and memory management
    • TT has more new String API improvements
  • This Week's Status:
    • merged m-c and TT latest
    • adding tracing for various opcodes
    • TT prepping for SSE opts
    • landed nanojit refactorings

Landed some incremental perf wins, 5-10% improvement over 1.9 so far.

• Need to prioritize/file these.
  • JSObject ops makeover
  • direct threaded interpreter loop on Windows
  • SDK / compiler adjustments on Mac
  • Profile guided input improvements

Mobile 1.9.1 Update

• DirectFB port
  • Code in good shape, needs a final review and to be landed on mozilla-central
• bug 437948 Geolocation
  • Dougt has a patch to add geolocation support to the DOM.
• bug 436083 Viewport meta tag support
  • Bobby has a patch together than hooks in to Fennec to support Apple's proposed viewport meta tag

General 1.9.1 Updates

• offline, dcamp:
  • Last week:
    • versioned app caches just about ready for review
    • honzab has opportunistic caching namespaces almost ready
    • need to review localStorage patch
  • This week:
    • Versioned app cache patch attached 442806 (with some followups in the dependencies). Waiting for review.
    • Opportunistic/fallback caching namespace patch attached 442813. One round of review done, waiting on a new patch.
    • Test suite started 443017. Some review done, will need an updated patch.
    • Image requests/caching will be an issue, see bug 442809
• GSoc Windows Text Services Framework
  • Seems to be making significant progress without much help:

http://wiki.mozilla.org/Community:SummerOfCode08:TSF

mozilla-central

• Need a query facility, like bonsai. Status?
  • Need pushlog code updated on hg.mozilla.org to use new sqlite backend bug 437637, then pushlog web extension needs query support added bug 437445
• Cycle times are seemingly slower on moz-central (about an hour slower). Need to look into this. John O to investigate. Sorry, no chance to investigate yet. (01july2008)