[vlc-devel] [RFC PATCH 8/9] RFC: es_out: fix group_id and i_id collision

Thomas Guillem thomas at gllm.fr
Fri Feb 14 10:42:14 CET 2020



On Thu, Feb 13, 2020, at 19:01, Rémi Denis-Courmont wrote:
> Le torstaina 13. helmikuuta 2020, 13.03.37 EET Thomas Guillem a écrit :
> > Now the big question, do we want to do the same with libvlc ?
> 
> Of course, we do. LibVLC should not have a worse interface, should it?
> 
> We still lack a synchronous interface, that would be better for programmatic 
> (non-interactive) use cases, but that's a separate problem also affecting the 
> core.

What do you mean with programmatic interface ? Is that what I'm trying to do ? Having a way to select a track via a string, before the player is even started.

> 
> > (...) But I still want to be able to select a track by its string_id.
> > Mainly to restore the player previous state from a db.
> 
> For that, you need a reproducible serial identifier. The string identifier in 
> this patch is *no* such thing. It is a serial but unreproducible identifier. It 
> might work for RPC use cases like HTTP and D-Bus controls, but it won't work 
> for replaying the same MRL.
> 
> There are several sources of non-determinism between the MRL and the ES output 
> that would need to be resolved first.

Is it possible to tell me which one ?

If I understand correctly, there can be some case where tracks are not added in the same order, and it will impact the id order.

A lot of demuxers are adding track in the open callback(). In that case, the order is always the same, do we agree ? 

> 
> -- 
> レミ・デニ-クールモン
> http://www.remlab.net/
> 
> 
> 
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel


More information about the vlc-devel mailing list