On 22/02/12 11:48, Martin Storsjö wrote:
> On Wed, 22 Feb 2012, Luca Barbato wrote:
> 
>> On 22/02/12 10:30, Martin Storsjö wrote:
>>> FFABS(dts - times[1]*1000) > 5000
>>
>> Why that?
> 
> The times[] array is expressed as integer seconds, so the index
> timestamp won't be an exact match with the timestamp in the packet. And
> by using a slightly larger margin than just 500 ms, we wouldn't throw
> away an otherwise-good index if it's slightly off.
> 
> But Vladimir wasn't too keen on doing the validation on startup each
> time (since it will cause a slowdown on startup for all valid files),
> and instead suggested trying to validate it when seeking instead - I'll
> try to look into that to see if it's doable.
> 
> // Martin
> 

My problem was the magic numbers, I do agree that we should try to
consider the decoder/demux init not a slow path.

lu

-- 

Luca Barbato
Gentoo/linux
http://dev.gentoo.org/~lu_zero

_______________________________________________
libav-devel mailing list
[email protected]
https://lists.libav.org/mailman/listinfo/libav-devel

Reply via email to