[vlc-devel] Re: vlc: svn commit r8350 (gbazin)

[Gildas Bazin]

On Monday 02 August 2004 13:26, Mark Moriarty wrote:
> Question -- Is it reasonable to make the changes I described Friday,
> changing input_decoderdecode to an int and passing the error state (FIFO
> filled) back upstream?  That provides a hook for the calling function to
> take other action, whether a return 0/kill the Demux or a parameter change
> (reduce framerate for Screen: perhaps).
> 

I'm not really sure about this, I guess it could be a good idea but we 
really need to have a better fifo overflow detection (based on time and not 
on size). I guess I'll apply part of the patch though.

>
> I've been playing with the getthreadtimes function, which allows 
measurement
> of CPU use.  It might be interesting to use during initialization,
> auto-selecting a framerate which keeps CPU use at roughly XX percent.  The
> function doesn't apply to Win95/Win98, unfortunately.  There's a similar
> call in Linux, though it sounds like the Linux graphics approach doesn't
> have the issue Windows does.
> 

I actually like this idea because it can be all integrated in the screen 
plugin itself without touching any of the core of VLC. Well, at least as a 
first step. If this works out well we could maybe find a way to integrate 
it into the core at a later point (Although I'd be more inclined in having 
a frame rate control in the transcoder module itself).

--
Gildas

-- 
This is the vlc-devel mailing-list, see http://www.videolan.org/vlc/
To unsubscribe, please read http://developers.videolan.org/lists.html
If you are in trouble, please contact <postmaster at videolan.org>