[FFmpeg-devel] lavd API discussion

[Paul B Mahol]

On 9/13/16, Josh de Kock <josh at itanimul.li> wrote:
> Hi all,
>
> So I know that not as many people use lavd, but I just wanted to ask a
> couple questions about the lavd/lavf APIs (but mostly about lavd), maybe
> fuel some thought about potentially developing/evolving the lavd API
> further.
>
> 1. What is the real difference between lavf and lavd?
>
> Both lavf and lavd use AVInputFormat and AVOutputFormat, and from what I
> can see, there is no real difference from lavd/lavf apart from their
> folder names, and the general types of formats which they contain.
>
> 2. Does lavd need to be extended?
>
> So I can see where the original idea may have come from for
> lavd--something which can interface with devices, similar to lavf but
> not quite as 'passive'. Except it stayed as 'passive' as lavf, whereas
> an 'active' lavd may have been able to take commands and interface with
> devices with I/O as one might need. For instance with something like a
> camera, you may need to pause, zoom-in, zoom-out, etc. Or an old tape
> camcorder could take rewind, or fast-forward commands. And I'm sure
> there are a lot more instances you can think of where input or output to
> a 'device' outside of video, audio, subtitles, etc would be useful.
>
> If I'm mistaken, then please point it out, I'm still learning about the
> code-base. (And if this discussion has happened before, please link me
> to it, I couldn't find anything, but I may have just been searching for
> the wrong thing).

Some people dislike lavd, mostly because thay do not need/use it.

I'm all for improving API and making it more useful, so it can be reused
by ffplay for example.