[MPlayer-cvslog] r18780 - trunk/mplayer.c

[Uoti Urpala]

On Wed, 2006-06-21 at 22:29 +0200, michael wrote:
> Log:
> send flip_page() through the filter chain
> ignoring the filter chain cannot and does not work with frame skip / insertion filters
> if this breaks something then just reverse it, ive not exhaustively tested it ...

I think this is wrong, even though it might not break anything now. At
least it's not the right way to do what you implemented for vf_framestep
- the basic frame skip or insertion functionality cannot be based on
flip_page(). If a frame is skipped the main loop should know about it
before sleeping to time it. If a new frame is inserted by a filter the
main loop also needs to know about that so it won't feed more data to
the demuxer yet.

Does this even do anything useful with vf_framestep? There's already the
blit_page return value from the filters, shouldn't that already do
basically the same thing as the interception of FLIP_PAGE you added?