[FFmpeg-cvslog] r16363 - trunk/libavformat/sierravmd.c

[Mike Melanson]

Kostya wrote:
> On Sun, Dec 28, 2008 at 03:35:08PM +0000, Robert Swain wrote:
>> 2008/12/27 kostya <subversion at mplayerhq.hu>:
>>> Author: kostya
>>> Date: Sat Dec 27 18:34:00 2008
>>> New Revision: 16363
>>>
>>> Log:
>>> Latest Coktel Vision VMDs contained Indeo 3, add demuxer support for it
>>>
>>> Modified:
>>>   trunk/libavformat/sierravmd.c
>>>
>>> Modified: trunk/libavformat/sierravmd.c
>>> ==============================================================================
>>> --- trunk/libavformat/sierravmd.c       Sat Dec 27 17:25:36 2008        (r16362)
>>> +++ trunk/libavformat/sierravmd.c       Sat Dec 27 18:34:00 2008        (r16363)
>>> @@ -49,6 +49,7 @@ typedef struct VmdDemuxContext {
>>>     unsigned int frames_per_block;
>>>     vmd_frame *frame_table;
>>>     unsigned int current_frame;
>>> +    int is_indeo3;
>>>
>>>     int sample_rate;
>>>     int64_t audio_sample_counter;
>>> @@ -91,6 +92,10 @@ static int vmd_read_header(AVFormatConte
>>>     if (get_buffer(pb, vmd->vmd_header, VMD_HEADER_SIZE) != VMD_HEADER_SIZE)
>>>         return AVERROR(EIO);
>>>
>>> +    if(vmd->vmd_header[16] == 'i' && vmd->vmd_header[17] == 'v' && vmd->vmd_header[18] == '3')
>>> +        vmd->is_indeo3 = 1;
>>> +    else
>>> +        vmd->is_indeo3 = 1;
>> So vmd->is_indeo3 == 1 always? Was that intentional?
> 
> Of course not. Thanks for spotting (since there are only two people
> who care about VMD and the other one even has not added it to Fate
> tests suite, it could be long before finding it out).

According to my notes, I had trouble getting the VMD system to work when 
I was first trying to add a FATE test, which would make such a test 
meaningless. I will investigate it again soon to see if the situation 
improves. I suspect we will need several VMD tests for different 
generations of files.

-- 
     -Mike Melanson