On 9/29/12 11:07 PM, Luca Barbato wrote:
Now with the main refactor split in two patches.
I hope to get to push the first 3 patches soon, the 4th requires
maybe a little more discussion on what to do regarding the new api
and if we want to use the avformat_mux_ namespace for all the
related functions.
More rationale on why I want to split allocation/init and the actual write.
For segmenting you ideally get to configure your muxer format and then
you may or may not write the header more than once and write the trailer
more than once.
When you write a streaming server ideally you'd like to start streaming
as quick as possible once a client starts asking so, depending on the
format, you have something like this
init the muxer once
get a request -> write the header
-> write the next packet
The whole "write the header" operation should be as short as possible
time-wise, bonus point if the whole muxer structure could be shared
across, changing just the avio used.
I hope that helps a bit.
_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel
