[vlc-devel] [PATCH 0/8] Add drain support in sout

[Thomas Guillem]

I tried but it's very hard to do without these commits.

I tried to delete streams and sout to drain but then we can't seek anymore. Very problematic since the buffer can be 10seconds long. I'll really need to drain like decoders.

On Tue, Jan 30, 2018, at 09:38, Thomas Guillem wrote:
> 
> 
> On Mon, Jan 29, 2018, at 19:26, Rémi Denis-Courmont wrote:
> > Le maanantaina 29. tammikuuta 2018, 19.25.48 EET Thomas Guillem a écrit :
> > > Hello,
> > > 
> > > This set of patches modifies the sout_mux/sout_access/sout behavior by
> > > handling drain when a NULL buffer is sent to pf_write/pf_send callbacks.
> > 
> > IMO, draining only makes sense on single ES. Furthermore, draining is only 
> > meaningful in contrast to flushing. In stream output, you can assume that an 
> > ES is being drained when it gets deleted. So there is really no point in 
> > adding explicit draining, AFAICT.
> 
> From my cc sout, I can drain the next sout chain by deleting them, I 
> agree. But how do I know that I should wait for the client to read the 
> final http stream when I'm closing ?
> 
> That's why a drain is more convenient. Finally I would only need 3/8, 
> 5/8 and 8/8 patches to achieve my goal. Otherwise, I don't really know 
> what to do.
> 
> > 
> > And then, the generic problem of deadlock muxers can only be fixed by passing 
> > the PCR to the stream output - which is far more powerful and generic than 
> > draining.
> > 
> > -- 
> > 雷米‧德尼-库尔蒙
> > https://www.remlab.net/
> > 
> > _______________________________________________
> > vlc-devel mailing list
> > To unsubscribe or modify your subscription options:
> > https://mailman.videolan.org/listinfo/vlc-devel
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> https://mailman.videolan.org/listinfo/vlc-devel