[FFmpeg-devel] [PATCH 2/2] ffprobe: add -o option

[Marton Balint]

On Sun, 3 Apr 2022, Stefano Sabatini wrote:

> On date Wednesday 2021-04-21 23:57:04 +0200, Stefano Sabatini wrote:
>> On date Monday 2021-04-19 11:26:49 +0200, Michael Niedermayer wrote:
>>> On Sun, Apr 18, 2021 at 11:30:58PM +0200, Stefano Sabatini wrote:
>>>> This enables printing to a resource specified with -o OUTPUT.
>>>>
>>>> Address issue: http://trac.ffmpeg.org/ticket/8024
>>>> ---
>>>>  doc/ffprobe.texi  |   7 ++
>>>>  fftools/ffprobe.c | 174 ++++++++++++++++++++++++++++++----------------
>>>>  2 files changed, 120 insertions(+), 61 deletions(-)
>>>
>>> This seems breaking some fate tests like fate-gaplessenc-itunes-to-ipod-aac
>>>
>>> I see 0 bytes in the output:
>>>   73 3D 4B 44  00 73 69 64  65 5F 64 61  74 61 7C 0A  0A
>>>
>>> thx
>>
>> That's right, I'm still forgetting to run make fate :-)
>> (avio_put_str() was adding the 0), should be fixed now.
>>
>> Thanks.
>
> Updated again, now it's locally passing fate with FATE samples (don't
> remember if other issues were spotted the past time).

This looks good to me in general, so I intend to apply. One thing that we 
might do is to keep writing to the standard output if no output file is 
specified instead of using the special "pipe:" URL. This way ffprobe 
keeps working even if the pipe protocol is not compiled into 
ffmpeg/libavformat.

Regards,
Marton