[FFmpeg-devel] [PATCH 1/9] Allow reading of growing avi files (ie currently being written)

Joakim Plate elupus at ecce.se
Sun Sep 11 15:41:19 CEST 2011


On Sat, 10 Sep 2011 22:09:21 +0200, Michael Niedermayer wrote:

> On Sat, Sep 10, 2011 at 07:22:44PM +0200, Joakim Plate wrote:
>> [11 quoted lines suppressed]
> 
> this breaks several fate tests, thus not pushed yet
> please check if this is intended or a
> bug. If its intended please explain why it happens and include the
> checksum updates in the patch
> thanks
> 
> --- ./tests/ref/fate/truemotion1-24    2011-09-09 22:56:47.879195242 +0200
> +++ tests/data/fate/truemotion1-24  2011-09-10 22:06:04.259229514 +0200
> @@ -30,6 +30,7 @@
>  1, 82898, 10832, 0x3c746fbe
>  0, 84000, 69120, 0x5876d758
>  1, 88424, 10832, 0x05f3b08a
> +0, 90000, 69120, 0x45e7dd5c
>  1, 93951, 10832, 0xa6560483
>  1, 99478, 10832, 0xd98a8e19
>  1, 105004, 10832, 0xf98a0b2e
> 
> 
> [...]

Okey, looked at this a bit. The issue is that the fate samples all have
riff headers for the original file, but then they have been bytewise cut
shorter. So the old code never output the last frame that has been cut off
somewhere in the middle. After the change that half packet is output.

>From what I gather you prefer if packets and frames are output even if not
complete. So i'll fixup the fate refs. 

/Joakim



More information about the ffmpeg-devel mailing list