TPEPlatform/2013-12-12

From MozillaWiki
Jump to: navigation, search

Media

Media Recording

  • bug 879668 - [MediaEncoder] Implement OmxTrackEncoder on B2G / John Lin
    • revise patch to address review comments and upload again.
    • study Mozilla GTest tests.
  • bug 879669 - Support Video Encoder module in MediaEncoder framework. / Randy
    • Got r+ for media encoder and I would open another to land it.
  • bug 891705 - [MediaEncoder] Implement WebM 1.0 container writer. / Randy
    • Can link to libmkv and do embl compositor, The result data can be parsed but video fps seems wrong.
  • bug 891704 - [MediaEncoder] Implement MP4Writer / ayang
    • Cowork with John for AAC/AVC track encoder problem
    • Cowork with Randy for MediaEncoder A/V recording path problem
    • Checking output muxing content format.
  • bug 945127 - OpusTrackEncoder should correct the checking statement of source / Shelly
    • review+, landed.
  • bug 948819 - [MediaEncoder] Remove the current implementation of rounding odd size video resolutions / Shelly
    • review+, landed.
  • bug 948815 - [MediaEncoder] Handle odd size video resolutions / Shelly
    • Save for future discussion and implementation.


WebAudio

  • bug 916399 - Intermittent test_timeupdate_small_files.html | Test timed out / JW
    • run tests on try server
    • review code of related bug: Bug 860760, Bug 872462
    • buffer timeout caused by broken implementation of OMXCodec::pause
    • calling OMXCodec::start won't leave paused state and return error
  • openh264 / JW
    • trace code, run test program
    • discucss with Ekr, write unit tests with GTest

Streaming Media

GetUserMedia

  • bug 898949 - [B2G getUserMedia] Display front/back camera list on permission prompt / SC
    • r+ from @fabrice, but need test case for it.
    • some more work need to be done for contact permission (PermissionPromptHelper/PermissionPromptService)
  • bug 947010 - Create automation test for recording status / SC
    • survey our existing test framework, will use marionette for it.
  • bug 853356 - [B2G getUserMedia] Display camera/ microphone permission acquisition prompt / ayang
    • co-work with Randel for bug 945614
    • it needs to add testcases for gUM unknown, contact/device storage and other no test permission prompt
  • bug 926746 - HARDWARE_UNAVAILABLE is already thrown while GetUserMedia for audio / slee
    • landed
  • bug 946562 - gUM performance on emulator / slee
    • trace emulator kernel
  • Media pipeline time consumption profiling / slee

Graphics

  • bug 937713 - [B2G] Display does not respond or responds too slowly to device rotation. /vilin
    • Main thread didn't send update to Compositor. Investigating why nsWindow:onDraw() is not called.
  • bug 946952 - Layer tree during Camera preview is inconsistent. /vilin
    • Background layer sometimes will be ColorLayer or ThebesLayer
    • Make Image Layer fit to display resolution and see if background layer will be removed.
  • bug 947504 - [v1.3] Screen colors are off and display blinks /vilin
    • HWComposer issue. Lack of HWComposer patch in v1.3 code.
  • bug 944564 - [B2G] Page transition in Homescreen can't reach 60fps stably. /vilin
    • Breaking down 3 problems cause this.
      • ProcessRestyles() is missed(not called) randomly even touch event is received and tick is fired.
        • bug 862276 - reduce costs from descendants of simple dynamic style changes (e.g., changing 'transform' in style attribute)
      • Input event sometimes being queued in B2G main thread when B2G is busy.(e.g GC is ongoing) Then it can't be transmitted to Homescreen in time.
        • bug 930939 - (input-thread) Separate the threads that run gecko and that receive input events from the OS
        • Take advantage of InputManager ?
          • "InputRead" thread takes care of event receiving ; "InputDispatcher" thread takes care of event dispatch
          • An option, but dispatch touch event() will run gecko code(reflow). How to decouple event dispatch and reflow ?
      • Timer for nsRefreshDriver sometimes lags a lot.
        • Need more investigation ...
  • bug 941984 - [FlatFish] InputReader takes ~18ms to get each event. Makes FPS bounded at ~50fps. /vilin
    • Device-specific problem in driver(also happened when it's running on Android). Partner is sitll investigating...
  • bug 893304 - Convert CanvasClientWebGL (and Host) to the new textures /Morris
    • Resolved some test failure on try server
    • Wait for review
  • bug 931373 - Optimize PROFILER_MAIN_THREAD_LABEL /Jerry
    • Avoid to use TLS when we are in main thread
  • bug 945992 - Crash in CompositorChild::ActorDestroy while receiving calls and SMS continuously /Jerry
    • B2G will raise abort when IPC broken.
      • Wait for dmesg/logcat log
  • bug 942199 - Camera filmstrip preview shows partially loaded snapshot for half a second /Jerry
    • Nexus4's camera prview size is small than the screen resolution. Thus, camera app will load the full-size image from storage.
  • bug 929432 - [B2G][Helix][Browser][ChenHoulai] Open youtube in the browser, there is an extra horizontal line on the top when in full-screen mode /Peter
    • wait for review
  • bug 925444 - Gecko layers should honor hwc releaseFenceFd /Peter
    • Debug with HWC integration
  • bug 946596 - Hi-res phone does not take hi-res screenshots
    • Debugging
  • bug 946245 - [Display][gonk-kk] Porting GonkDisplay, nativewindow and libui / Solomon
    • Fix GonkDisplayJB.cpp by coupling the relative display component(FramebufferSurface, BufferQueue, Surface) properly according KK's change.
    • B2G still crashes, debugging with components of LayerTransaction(parent, child, and relative IPC calls). The root cause probably orginate from Gralloc initialization.

Input

Stability

Free Agent

  • TaskTracer / Shelly
    • Brown bad sharing about TaskTracer
    • TODO: Add external source event.

Study

  • bug 925615 - Support Wifi Display /vichen
    • Study WFD related Spec
    • Trace Gecko Layout
  • imageLib / SC
    • loading/decoding/rendering process
    • MemShrink facilities we already have

bird-eye's view of imageLib