revid: move ringbuffer closer to start of pipeline
I propose that the ringBuffer be moved closer to the start of the pipeline; placed after lexing. This way, we can have different methods of packetization and therefore two different outputs requiring differing packetization methods. This is particularly attractive for dual rtmp and rtp streaming, which is desired for use with the sled.
On first inspection, this will require some sort of routine that will read from the ringbuffer and send h264 nal units to the encoders we wish to use. It also means that the design of the senders will change, in particular, no more loading of ringBuffer chunks, it may even be that the load methods become redundant. It might be appropriate to make the the senders implement io.Writer.
Comments (3)
-
reporter -
reporter more details in this doc here
-
reporter - changed status to resolved
addressed in PR #167
- Log in to comment
I will create a design doc once I have a better idea on how I want this to work. but please go ahead and create one if you already have some ideas.