Issue metadata
Sign in to add a comment
|
Allow running demuxers and all decoders in a Utility Process
Reported by
julien.i...@gmail.com,
Jan 17 2017
|
||||||||||||||||||||||
Issue description
UserAgent: Mozilla/5.0 (X11; Ubuntu; Linux x86_64; rv:40.0) Gecko/20100101 Firefox/40.0
Steps to reproduce the problem:
HTML5 progressive streaming and MSE.
What is the expected behavior?
What went wrong?
- VideoFramePool does not work with MojoSharedBufferVideoFrame which prevents FFmpeg/Vpx/VideoDecoder to fill mojo video frames.
- MojoRender does not start the video_sink and the audio_sink. It is currently limited to overlay.
- There is no proper UtilityMojoMediaClient. Currently the test mojo client is created for utility process.
- Demuxers are not mojified in the sense that it is not possible to run ChunkDemuxer and FFmpegDemuxer into a separate process than the renderer process.
Did this work before? N/A
Chrome version: 57.0.2985.0 (Developer Build) (64-bit) Channel: dev
OS Version:
Flash Version: (Disabled)
The work has been splited into 3 steps:
1- Allow to use {FFmpeg/Vpx}VideoDecoder from a UtilityProcess:
enable_mojo_media = true
mojo_media_services = [ "video_decoder" ]
mojo_media_host = "utility"
2- Allow to use the DefaultRendererFactory from a Utility Process:
enable_mojo_media = true
mojo_media_services = [ "renderer" ]
mojo_media_host = "utility"
3- Mojify demuxers and allow running {Chunk/FFmpeg}Demuxer in a Utility Process:
enable_mojo_media = true
mojo_media_services = [ "demuxer", "renderer" ]
mojo_media_host = "utility"
1 and 2 are described here: https://docs.google.com/document/d/19n22cLElIoBEBepXA4TslHfVWRbzf37eThzppmXgTac/edit#
3 is described here: https://docs.google.com/document/d/1PLuRatPNRHd7T2osFL7gNxgKtGLW8vyYidgfOZDvBFE/edit#
Since I was not familiar with mojo and so the media mojification I did the design as I was implemented it. I had several rounds of refine so switching between design and implementation all along this run. So that is why I am coming with an implementation already which is not just a proof of concept. But let's start from the beginning by discussing the design first. I am open to any change.
,
Jan 17 2017
> I do not know how to teach git cl upload to avoid a CL to contain all > previous CLs. git cl upstream <branch name> git cl upload
,
Jan 17 2017
,
Jan 18 2017
> git cl upstream <branch name> > git cl upload Thx Dan for the tips. I re-submitted the CLs using that. Also I splited this issue into 2 so I am closing this one. |
|||||||||||||||||||||||
►
Sign in to add a comment |
|||||||||||||||||||||||
Comment 1 by julien.i...@gmail.com
, Jan 17 2017