[FFmpeg-devel] [Ffmpeg-devel] [PATCH] Debug help for file position

[Benoit Fouet]

Michael Niedermayer wrote:
> Hi
>
> On Wed, May 02, 2007 at 12:08:05PM +0200, Benoit Fouet wrote:
>   
>> Michael Niedermayer wrote:
>>     
>>> Hi
>>>
>>> On Wed, May 02, 2007 at 11:18:09AM +0200, Benoit Fouet wrote:
>>>   
>>>       
>>>> Michael Niedermayer wrote:
>>>>     
>>>>         
>>>>> Hi
>>>>>
>>>>> On Wed, May 02, 2007 at 10:02:49AM +0200, Benoit Fouet wrote:
>>>>>   
>>>>>       
>>>>>           
>>>>>> Michael Niedermayer wrote:
>>>>>>     
>>>>>>         
>>>>>>             
>>>>>>> Hi
>>>>>>>
>>>>>>> On Mon, Apr 30, 2007 at 05:42:06PM -0700, Daniel Cardenas wrote:
>>>>>>>   
>>>>>>>       
>>>>>>>           
>>>>>>>               
>>>>>>>> In utils.c, function av_read_frame_internal, the position offset in the 
>>>>>>>> file is not passed along in the packet.  This one line change helps with 
>>>>>>>> that issue.  Helpful when debugging issues with file based content.  Its 
>>>>>>>> not a perfect fix. Not sure how the packet/buffer management works so the 
>>>>>>>> position isn't quite right, but close.  If you guys like this patch, then 
>>>>>>>> perhaps I'll figure out the perfect fix and submit that.  Again, this just 
>>>>>>>> helps with debug.
>>>>>>>>     
>>>>>>>>         
>>>>>>>>             
>>>>>>>>                 
>>>>>>> iam fine with the patch if it doesnt break regression tests
>>>>>>>
>>>>>>> [...]
>>>>>>>   
>>>>>>>       
>>>>>>>           
>>>>>>>               
>>>>>> it breaks seek tests, but regression tests are ok
>>>>>>
>>>>>> outputs lots of things like:
>>>>>> -ret: 0 st: 1 dts:3.100000 pts:3.100000 pos:29375 size:208 flags:1
>>>>>> +ret: 0 st: 1 dts:3.100000 pts:3.100000 pos:-1 size:208 flags:1
>>>>>>     
>>>>>>         
>>>>>>             
>>>>> adding proper pos for the cases where it was -1 before is ok assuming
>>>>> there are no other changes in the output
>>>>>   
>>>>>       
>>>>>           
>>>> to be sure we understand each other
>>>> before the patch: pos = xxxxxx
>>>> after the patch:    pos = -1
>>>>
>>>> i guess you understood the opposite, no ?
>>>>     
>>>>         
>>> $grep 'ret: 0 st: 1 dts:3.100000 pts:3.100000 pos:-1 size:208 flags:1' *.ref|wc -l
>>> 8
>>>
>>> $grep 'ret: 0 st: 1 dts:3.100000 pts:3.100000 pos:29375 size:208 flags:1' *.ref|wc -l
>>> 0
>>>
>>> [...]
>>>   
>>>   
>>>       
>> ok i read the opposite :) (i guessed diff -u was called with files in
>> the opposite way)
>>     
>
> yes and that should be fixed ...
>
>
>   
patch attached

>> i attach the output of:
>> $ ./seek_test.sh seek.regression.ref > seek_test.diff
>> it's pos which changes most of the time, except once:
>>     
>
> as already said iam ok with the pos changes and the original patch
>
>
>   
ok, i'll deliver the patch.
i guess the regression ref file should be delivered in the same commit,
is it ok ?

>>  data/b-libav.ts
>> -ret: 0 st: 0 dts:-0.040000 pts:0.000000 pos:-1 size:24921 flags:1
>> +ret: 0 st: 0 dts:95443.677689 pts:0.000000 pos:-1 size:24921 flags:1
>>
>> but this is not linked to this patch
>>     
>
> hmm, and what caused that?
>
> [...]
>   
>   
dunno, i'm going to look after a revision in which seek tests worked.

Ben
-- 
Purple Labs S.A.
www.purplelabs.com
-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: seek_test.sh.diff
URL: <http://lists.mplayerhq.hu/pipermail/ffmpeg-devel/attachments/20070502/251168c6/attachment.asc>