TPEPlatform/2013-07-11: Difference between revisions
Jump to navigation
Jump to search
Line 62: | Line 62: | ||
==Meeting Minutes== | ==Meeting Minutes== | ||
CJ | ===CJ=== | ||
*On going event: | *On going event: | ||
**[https://taiwan.etherpad.mozilla.org/251 Gfx dairy sprint] | **[https://taiwan.etherpad.mozilla.org/251 Gfx dairy sprint] | ||
Line 71: | Line 71: | ||
**[https://taiwan.etherpad.mozilla.org/254 WebRTC weekly meeting] | **[https://taiwan.etherpad.mozilla.org/254 WebRTC weekly meeting] | ||
***10:00 AM each Friday. Blue mountain. | ***10:00 AM each Friday. Blue mountain. | ||
*Backlog | |||
**https://intranet.mozilla.org/TPEPlatform | |||
**Share your idea, what you want to do in the future. |
Revision as of 06:48, 11 July 2013
Media
Media Recorder
- Write document for B2G HW decode diagram / Alfredo
- Discuss architecture design on #media, studying mp4v2 as muxer / Alfredo
Streaming Media
RTSP bug 831645 / Bechen
- RTSP bug 831645
- waiting for review
- bug 877116 - [RTSP] audio playback latency is serious with the following media streams(Video:H.264 Audio:AAC)
- According to rfc3640, we can divide a RTP packet into several Access-Unit, and calculate the timestamp for each AU.
- It works now, but since we divide a RTP packet into several small pieces, also increase the ipc loading. And we observe the smallest aac AU is only 9 bytes (silence aac frame), means that a RTP packet will be divided into more than 100 AUs corresponding to 100 ipc, 100 slots. Then we must adjust the slot buffer number in RTSPTrackBuffer bug 877193.
- According to rfc3640, we can divide a RTP packet into several Access-Unit, and calculate the timestamp for each AU.
- The b2g process crash on our ipc call "onMediaDataAvailable", seems we can't do ipc on android thread. Help by / Cervantes.
WebRTC
- bug 869869 - e10s for UDP socket / SC
- update patch according to 4th review comment
- investigating memory leakage in nsUDPSocket, leak at DeadlockDetector (on exist on debug build)
- performance analysis / Steven
- Huawei device performance test
- Test ekr's video benchmark
- Try to use perf from jld
- It has problem when parsing the recorded data, asking jld.
- memcpy from GraphicBuffer to system memory is slow.
- Still trying to find out why it is so slow(copy 640x480 bytes takes about 12ms)
Graphic
Canvas 2D/SkiaGL bug 858237
- bug 869199 Bad performance of fishie tank/ Peter
- Implement separate thread for canvas rendering and check bug 857895
- bug 875168 Integrate SurfaceTexture/ Peter
- pending...
HW Composer
- bug 884188/bug 885345 HWC can't render small layer due to no gralloc buffer/ Peter
- Provided workaround on m-c and wait for review
Misc
- Prepare b2g GFX introduction and porting([1])
- Have presentation at 10:30 am this Friday
Stability
Free Agent
B2G GFX arch
- Reading GFX arch /JerryShih
- Studying Gfx/Layer architecture /Morris
MMS
- bug 889767 - [MMS] AMR audio attachement is recognized and shown as video file \ SC
- add attachment of a *.3gp audio file will be recognized as a video blob
- nsExternalHelper return video/3gpp for *.3gp file, change the file name to *.amr will correctly return audio/amr.
- *.3gp is a container format, should allow *.3ga for audio file instead.
- bug 887164 - [MMS-Gallery] BMP and WBMP display support from MMS issue. \ SC
- discuss with Steve and leo partner, decide to decode WBMP in Message app and save WBMP as PNG.
MemShrink
- bug 771765 - [Nuwa] update patches per review comments \Cervantes
- Awaiting review for the IPC-related parts.
- Study browser-element test cases.
Meeting Minutes
CJ
- On going event:
- Gfx dairy sprint
- Bring up gfx new onboards gfx module knowledge in Gecko
- Everyday. Host at gfx cubic
- Media weekly meeting
- 3:00 PM each Wednesday. Blue mountain.
- WebRTC weekly meeting
- 10:00 AM each Friday. Blue mountain.
- Gfx dairy sprint
- Backlog
- https://intranet.mozilla.org/TPEPlatform
- Share your idea, what you want to do in the future.