[FFmpeg-devel] [PATCH 0/5] Cuvid/videotoolbox preparations

[James Almer]

On 10/27/2017 3:46 PM, Thilo Borgmann wrote:
> Am 27.10.17 um 19:00 schrieb James Almer:
>> On 10/13/2017 1:59 PM, wm4 wrote:
>>> These commits are required to merge Libav's cuvid hwaccel, and to
>>> fix videotoolbox operation if frame threading is enabled.
>>>
>>> Anton Khirnov (4):
>>>   decode: avoid leaks on failure in ff_get_buffer()
>>>   decode: add a method for attaching lavc-internal data to frames
>>>   decode: add a mechanism for performing delayed processing on the
>>>     decoded frames
>>>   decode: add a per-frame private data for hwaccel use
>>>
>>> wm4 (1):
>>>   lavc/avrndec: remove AV_CODEC_CAP_DR1, as it's broken
>>>
>>>  libavcodec/avrndec.c         |   1 -
>>>  libavcodec/decode.c          | 113 ++++++++++++++++++++++++++++++++++++++++++-
>>>  libavcodec/decode.h          |  40 +++++++++++++++
>>>  libavcodec/h264dec.c         |   5 +-
>>>  libavcodec/huffyuvdec.c      |   3 +-
>>>  libavcodec/mpegutils.c       |   4 +-
>>>  libavcodec/vp3.c             |   3 +-
>>>  libavcodec/wrapped_avframe.c |   7 +++
>>>  8 files changed, 167 insertions(+), 9 deletions(-)
>>
>> Can we please find a way to get this thing moving instead of keeping it
>> blocked indefinitely? I'll eventually reach this point in the merge
>> queue and i don't want to find myself stuck again because two devs can't
>> find a common ground.
>>
>> We have dozens of devs, many knowledgeable enough in this subject to
>> chime in and tip the scales. So far one dev is against it (Michael) and
>> two in favor (wm4 and Thilo). This needs the attention of more people.
> 
> I'm not in favor of anything and never claimed to do so. Instead, I already said
> that I am not as familiar with that part of the code to dive into the discussion.
> I just offered to do the codemonkey on whatever shall finally be implemented -
> for the case it actually stalls for nobody willing to do so.

I recall you mentioned you were waiting for this to be committed (saying
you didn't think it was an issue like Michael thinks it is) so you could
go ahead with some hwaccel work you had queued.
Guess i wrongly assumed that meant you agreed with the approach. Sorry
about that.

> 
> 
>> If a vote is needed then so be it, but for fucks sake, we really need to
>> find solutions for this kind of discussions in a more timely manner.
> 
> Yes.
> 
> -Thilo
> _______________________________________________
> ffmpeg-devel mailing list
> ffmpeg-devel at ffmpeg.org
> http://ffmpeg.org/mailman/listinfo/ffmpeg-devel
>