[vlc-devel] [RFC PATCH 00/10] Activate chromecast live
remi at remlab.net
Thu Sep 7 16:49:25 CEST 2017
Le torstaina 7. syyskuuta 2017, 15.45.56 EEST Hugo Beauzée-Luyssen a écrit :
> As discussed during last workshop, this is an attempt at solving #18605 that
> doesn't invole a demux_filter.
> Instead, pausing, setting and fetching position/time is handled through new
> sout commands.
I don´t remember any such proposal. What I remember is that:
1) Stream output should be notified of PCR (via control request from ES
output). However this might not work until clock and synchronization are moved
to outputs. This is not related to renderers as such.
2) Demux filter could be enabled/disabled (*not* inserted/removed) dynamically
via a demux control from the input thread. However it will probably not work
properly if there are more than one chained demux filter.
3) The current display time should be estimated by querying the actual PTS
from the ES output, and underneath by the stream output, rather than by adding
the PTS delay. As with the PCR, this might be problematic until the clock and
synchronization are moved downstream.
Also this is really _not_ the same thing as getting the input position and
time. The stream output neither knows nor cares about those. Conversion
between PTS/PCR and input time/position can only be done by the demuxer or the
I) I don´t remember anything about adding pause/resume controls to stream
output. I don´t have a problem with the concept as such, but it sounds tricky
to implement properly with asynchronous demuxers and reentrant ES output.
II) I also don´t remember any discussion on setting the time to the stream
output. And in fact, I don´t really understand what that is even supposed to
mean, as noted already, the stream output has no clue about time/position.
More information about the vlc-devel