[vlc-devel] Matroska, libopus and loophole in the spec

[Timothy B. Terriberry]

Denis Charmet wrote:
>> Better would be to have the demux always provide this duration,
>> wouldn't it? You need information from the MKV layer to handle
>> pre-skip and end-trimming correctly, anyway.
>
> Then you need to make this mandatory in the spec. The demux can only

The spec for the demuxer API? I'm confused. What do you think needs to 
be made mandatory where?

> provide informations set by the muxer. In any case libavcodec seems to
> work without knowing the size in sample of the coded block so it feels a
> little overkill to force that especially if the decoder can find it by
> itself.

Well, the duration of a packet can be determined from the packet itself. 
I'm just saying it makes sense for the demuxer to look at the packet and 
figure out what it is.

> Seek preroll and end trimming will be handled at the demux level
> provided for the latter that we have a duration to reduce.

Unless the packet is not a valid Opus audio packet, it will always have 
a duration.