[MPlayer-cvslog] r20765 - in trunk/libmpdemux: demuxer.c demuxer.h

[Nico Sabbi]

Uoti Urpala wrote:

>"0 otherwise" is questionable IMO as it's a valid position (though
>nothing seems to check that now).
>
>With return type "int" this is suitable for OSD use only (and you might
>want a more precise value for that too in some situations).
>  
>
yes, I'll unify to double all timing-related functions

>I think it's very questionable to make a demuxer-level function to query
>the current time. The demuxer packets could be buffered and the demuxer
>shouldn't know which part is being _displayed_ at the moment. If there's
>a better guess about the current time than the value that is now set as
>pts in the packets 
>
whether the current timing is more precise than demux timestamps
depends on the implementation of _GET_CURRENT_TIME control;
in the case of dvd it's absolutely not better (something better than 1 
second of accuracy)

>then either 1) the pts should be set to that value
>instead,
>
not the case

> or if the value is not suitable for pts use then 2) it should
>be placed in another field associated with the demuxed packets.
>  
>

why there should be a link between demux packets (demux-stream specific)
and playtime (mux/stream specific) ?
Don't forget that this function is intended to show on the OSD the value 
of the current
playtime, not to be used as a central synchronization clock