[vlc-devel] Accessing the demux Control at the interface level ? (re: RTSP trickplay)

Glen Gray slaine at slaine.org
Fri Dec 7 10:18:52 CET 2007

There seems to be some confusion of the reason for having the option.

As Ross says, RTSP already handles returning the scale that the server 
used, regardless of what you asked. This is not why the option is there.

The reason is to do with VLC and how to trigger the call to 
DEMUX_SET_SCALE. At the moment the interface for  Fastforward and Rewind 
(this is my understanding based on the rc interface module at least) 
triggers a short skip forward or backwards in the current play buffer by 
simulating a key-press event, which is different to setting the scale of 
play on an RTSP stream.

The option is there at the moment, to specifically allow for other 
methods of implementing FF/RW. My suggestion was to replace the user 
supplied option with a query to the Demux to see if it supported setting 
the scale, if so then enter that code path, if not then carry on with 
the existing FF/RW implementations.

There may well be a better implementation technique for this in VLC, 
suggestions welcome.

Jean-Paul Saman wrote:
> Ross Finlayson wrote:
>>> Laurent Aimar wrote:
>>>>   I will have a look at your patches this WE (or this week if I can find the
>>>>  time)
>>> The patch looks OK to me except for the option to turn scale on. That
>>> renders the whole usage of scale unusable for a simple user. There most
>>> by some way of either:
>>> 1) letting live555 libs handle it
>> The "live555 libs" (or, more precisely, the RTSP protocol) already 
>> handles this.  Regardless of what scale you specify in the RTSP 
>> "PLAY" request, the server will - in its RTSP "PLAY" response - tell 
>> the client the *actual* scale that it supports.  (For example, if you 
>> ask for a scale of 3.2, then the server might respond with a scale of 
>> 3, or a scale of 1 (if it doesn't support fast-forward at all).)  As 
>> I've noted already, after calling "playMediaSession()" (or 
>> "playMediaSubsession()"), simply call "session.scale()" (or 
>> "subsession.scale()") to get the actual scale factor that the server 
>> chose to support.
>> So, is there any reason not to *always* allow the user to (try to) 
>> change the scale?
> I think this is the *perfect* reason to do away with the proposed 
> options on the comandline for scale. Just assume the server will support 
> it and check the real value.
> Ross, Thanks for your clarification.
> Gtz,
> Jean-Paul Saman.
> _______________________________________________
> vlc-devel mailing list
> To unsubscribe or modify your subscription options:
> http://mailman.videolan.org/listinfo/vlc-devel

Glen Gray (slaine at slaine.org)

More information about the vlc-devel mailing list