[MPlayer-dev-eng] Lots of stuff for NUT

Michael Niedermayer michaelni at gmx.at
Wed Jan 4 19:38:54 CET 2006


Hi

On Wed, Jan 04, 2006 at 04:40:39PM +0200, Oded Shimon wrote:
> On Fri, Dec 30, 2005 at 09:10:02PM +0200, Oded Shimon wrote:
[...]
> Micheal:
> I sort of agree with your repeated index, that it should be allowed, but 
> only on one condition - it can only be immediately after the main headers 
> in the begginning of the file. Nowhere else except the very end of the file 
> is acceptable, as it causes the index altering itself. When put at 
> begginning of file, all offsets in the index need to be added to the length 
> of the index itself.

i disagree, leave this to the muxer just allow repeating, its not so hard
just find the "normal" index size during pass 1 add 5 or 10% space and 
reserve this during the second pass if it doesnt fit add another few percent
to whaever would have been needed and try again, or store all packet sizes&
timestamps and "simulate" the muxing after the first pass until a perfect fit
is found


> 
> I think I also agree that back_ptr and etc. can be made purely optional. 
> Nothing but the startcode and pts need to be written in "low percision" 
> mode, in which case when you seek you just seek to the syncpoint you want 
> and start playing, end of story. The index should also be as simple in low 
> percision mode. (A file's behavior should not change due to index)

agree


> 
> This way, for those who want can have low overhead and low percision for 
> their fun movies, and those who need video editing can have all the power 
> of active streams and etc. Default should certainely be high percision 
> though IMO.
> 
> Can I atleast commit steps 1 and 3? (2 is broken due to back_ptr thing).

ok

[...]

-- 
Michael




More information about the MPlayer-dev-eng mailing list