Platform/GFX/2010-Aug-9

index

• Every Monday at 2:30 PM Pacific/5:30 PM Eastern/21:30 UTC
• +1 650 903 0800 x92 Conf# 217
• +1 416 848 3114 x92 Conf# 217
• +1 800 707 2533 (pin 369) Conf# 217 (toll free, Skype)

Project Updates

• GFX
• Hardware Acceleration
  • The Path to Firefox 4 - what we need to get done over the next three weeks.
  • Current targets: Direct2D for beta 4, OpenGL & D3D9 accelerated layers for beta 5.
    • This is largely because beta 5 is feature freeze.
• Electrolysis

Status Updates

• Joe
  • Last week:
    • Short week due to civic holiday.
    • Fixed mobile bug 552809 with asyncDrawXULElement.
    • Coded and tested a probable fix to bug 583028. Looks good on try, has review, just need to find a time in which it's possible to land it.
    • Wrote Path to Firefox 4 document, referenced above.
    • Starting work on full browser accelerated layers with bug 580410 and bug 581213.
  • This week:
    • Damon's in Toronto.
    • Push for accelerated layers automated reftests, and fix bugs in accelerated layers.

• Jeff
  • Last week:
    • D2D reftests
      • Support for detecting d2d from reftests
      • Fuzzy matching prototype
      • Looked into some rasterization issues
  • This week:
    • D2D reftests (we might be getting very close this week)

• John
  • Last week:
    • Review of WOFF draft
    • Testing/review of TextRunWordCache crasher bug 553963
    • CSS3 Fonts edits
  • This week:
    • Post new version of CSS3 Fonts
    • Font crasher
    • Patch for Cleartype opacity bug 363861
    • Triage Windows/Linux font crashers

• Bobby
  • Last week:
    • Landed bug 343515 - Active Tab API
    • Switched to git!
    • Helped bz investigate bug 584092 - weird multipart/x-mixed-replace behavior
    • Finished up and landed bug 579122 - OnStartDecode race condition
    • Had quite a few long design discussions with dholbert on how to restructure imagelib
    • Support for jlebar on bug 573948 - Replacing libjpeg with libjpeg-turbo
    • Started investigating bug 585129 - Propogating mIsActive to resource document presshells
      • not sure if this should block - hopefully a driver knows?
    • Reviews for joe on bug 583028 - Tp4 regression fix
    • Started investigating bug 583825 - Progressive decoding no longer works
      • Have a decent handle on the problem, still not sure of the exact cause
    • Working on landing the tests for bug 343515
    • Big task - Worked on bug 512260 - Active presshell locking
      • CSS story turned out to be more complicated than I thought - lots of back and forth with dbaron
      • Patch stack up, ready for review
    • Reviewed all 11 patches in bug 584841
  • This week:
    • Sending psychic messages to dbaron to make him review my patch stack on bug 512260
    • After that, we should in theory be able to turn on discarding. It would be awesome if we could get it on for beta 4!
    • More work with dholbert
    • More investigation of bug 583825 (progressive decoding broken)
    • Next big task - bug 514033 - error recovery for imagelib (the source of most of my remaining blockers)

• Bas
  • Last week:
    • D2D reftest orange work
      • Custom operator support/cairo_d2d_device_t
      • Transformed stroke fixes
    • Some D2D memory diagnostics
    • D3D9 layers improvements
    • Minor CSS transform assistance
  • This week:
    • D2D reftest orange work
    • A little D3D9 layers improvements

• BenWa
  • Last week:
    • Exam studying.
    • Can't reproduce: bug 584488 - hang after attaching to external monitor.
    • reviews:
    • qcms
    • bug 578868 - [OOPP] OOPP are also loaded in main process
    • bug 577494 - Fullscreen in flash video does not hide dock, menubar or focus plugins windows.
  • This week:
    • bug 583296 - [Mac] Beach balling when I trigger the flash context menu
    • Any reproducible OOPP bugs.

• Benoit
  • Last week:
    • WebGL security meeting.
    • Big ongoing discussion about the mochitest in bug 582053.
    • Thought about how to handle the part of the GL ES spec that says that certain "bad" textures must be handled as 1x1 black textures. Considered using WebKit code there, but it's actually slow in the case that matters most, and not easy to extract, so, going to take a different approach.
    • bug 585199: Implement WebGLActiveInfo, adapt WebGL code to these changes. In particular, we don't need at all anymore to directly manipulate the JS context (except of course in custom quickstubs) so this patch removes Vlad's NativeJSContext helper class. However, this is blocked by what seems to be a bug in the JS engine on 64-bit platforms: bug 585714.
    • bug 584501: another round of texture fixes. Fixes validation of more parameters, fixes a bug with cube maps.
    • bug 584861: fix CheckSaneSubrectSize, wrt integer overflow and wrt zero sizes.
    • bug 584840: fix texParameter functions: validate parameters, discuss on the WebGL list what texParameterf should do and implement the outcome.
    • Patch reviews and bug triaging.
    • Set up the Toronto WebGL community mailing list, already 6 members!
  • This week:
    • Hopefully land all current patches, do the 1x1 black textures thing, and import the WebKit code to support fancy texture format conversions.

Infrequent / Irregular attendees

• Rob
  • Last week:
  • This week:

• Jim
  • Last week:
  • This week: