removing internal tremor (was: Re: [MPlayer-dev-eng] non-free and undistributable files in tremor/)

[Uoti Urpala]

On Fri, 2006-08-04 at 22:59 +0200, Balatoni Denes wrote:
> I support the removal of tremor (of course) but there could be one regression. 
> There is a hack in the ogg demuxer that asks tremor for the decoded size of 
> each frame. This is needed because in ogg not every audio packet has it's own 
> timestamp, but mplayer likes to always have the exact timestamp - so the 
> demuxer calculates the time delta from the decoded packet length. This is

It is not necessary for each audio packet to have a timestamp in normal
MPlayer use. MPlayer already does internally what you describe, counts
bytes since the last known timestamp (which it has to do anyway for byte
positions that are not at a packet boundary). However if there is no
timestamp in the first packets after a seek that can cause problems
since in that situation it's not possible to extrapolate from the
previous packets. Also encoding with streamcopy to another format might
require timestamps if the target container format cannot handle not
having a separate timestamp for every packet (though Mencoder timestamp
handling sucks so badly anyway at the moment that it's mostly a
hypothetical "if Mencoder was fixed otherwise..." question).

> c.) I don't remember what third alternative I had in mind.

Implement minimal vorbis parsing support in the ogg demuxer? (I don't
know how complicated it would be to get the packet lengths).