[FFmpeg-soc] [soc]: r4179 - in libavfilter: checkout.sh diffs/01_ffplay_filters.diff diffs/02_ffmpeg_filters.diff

[Vitor Sessak]

Stefano Sabatini wrote:
> On date Sunday 2009-03-22 19:10:34 +0100, Michael Niedermayer encoded:
>> On Sun, Mar 22, 2009 at 09:11:58AM +0100, Stefano Sabatini wrote:
>>> On date Sunday 2009-03-22 02:49:12 +0100, Michael Niedermayer encoded:
>>>> On Sun, Mar 22, 2009 at 01:08:24AM +0100, Stefano Sabatini wrote:
>>>>> On date Sunday 2009-03-22 01:05:42 +0100, stefano encoded:
>>>>>> Author: stefano
>>>>>> Date: Sun Mar 22 01:05:42 2009
>>>>>> New Revision: 4179
>>>>>>
>>>>>> Log:
>>>>>> Update to FFmpeg r18133.
>>>>>>
>>>>>> Modified:
>>>>>>    libavfilter/checkout.sh
>>>>>>    libavfilter/diffs/01_ffplay_filters.diff
>>>>>>    libavfilter/diffs/02_ffmpeg_filters.diff
>>>>> This breaks regression for mxf_d10.
>>>>>
>>>>> It is expected, since that test uses -padtop, which works differently
>>>>> from the one in FFmpeg (also the soc one looks broken).
>>>> he?!
>>>> there should be no -pad* in libavfilter svn because that should be done
>>>> by filters.
>>> Do you mean that the various -padXXX options should be removed from
>>> ffmpeg.c right now?
>> from soc libavfilter ffmpeg yes thats an option

I think ours users might hate that the command-line options they use is 
all of a sudden broken. I think it is reasonable having "-padxxx" 
supported and printing "Warning, parameter -padtop is deprecated, use 
instead -vfilters pad" during quite some time before removing that...

>>> Only problem is that regression test then will fail, so I suggest to
>>> wait up to when will have a pad filter.
>> id suggest to drop all pad options from the regression tests
> 
> So the question is why the mxf_d10 test uses -padtop in the first
> place.

It looks useful to me to detect if there is some regression on the 
padding code and it will be useful to test lavfi if the -padtop 
command-line switch is not removed...

-Vitor