Re: [FFmpeg-devel] [PATCH] libavfilter/vf_libvmaf.c The libvmaf filter tried to join on an invalid thread id

[James Almer]

On 5/3/2018 8:38 AM, Ronald S. Bultje wrote:
> Hi,
> 
> On Thu, May 3, 2018 at 4:41 AM, Kevin Wheatley 
> wrote:
> 
>> Following up my own email with another question or so:
>>
>> Could somebody point me at a suitable method of testing this within
>> the Fate framework?
> 
> 
> Why?
> 
> Your patch fixes a bug, I don't think we test bugs in fate, just features.
> 
> Patch LGTM.
> 
> Ronald

Pushed then.
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] libavfilter/vf_libvmaf.c The libvmaf filter tried to join on an invalid thread id

[Kevin Wheatley]

On Thu, May 3, 2018 at 12:38 PM, Ronald S. Bultje  wrote:
> Why?
>
> Your patch fixes a bug, I don't think we test bugs in fate, just features.

I am perhaps making an incorrect assumption that the code has a bug
because it is never tested and that by adding a test that simply tries
to use it, the vmaf code is more likely to work. I'm happy to not
write a test if that is preferred!

Kevin
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] libavfilter/vf_libvmaf.c The libvmaf filter tried to join on an invalid thread id

[Ronald S. Bultje]

Hi,

On Thu, May 3, 2018 at 4:41 AM, Kevin Wheatley 
wrote:

> Following up my own email with another question or so:
>
> Could somebody point me at a suitable method of testing this within
> the Fate framework?


Why?

Your patch fixes a bug, I don't think we test bugs in fate, just features.

Patch LGTM.

Ronald
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel

Re: [FFmpeg-devel] [PATCH] libavfilter/vf_libvmaf.c The libvmaf filter tried to join on an invalid thread id

[Kevin Wheatley]

Following up my own email with another question or so:

Could somebody point me at a suitable method of testing this within
the Fate framework?

I've started to try unpick how the fate tests are run, etc

What I need is a prototypical example which presumably would need to
factor in (what I figured out so far):

libvmaf is an optional component so the test should be too.
The test should probably live in tests/fate/filter-video.mak
The function (CMD) called should live in tests/fate-run.sh?

For the actual test I need two inputs an original and an encoded frame
(sequence is ok too), that can then feed into the vmaf invocation of
ffmpeg. Is there a set of inputs I can just reuse from what fate would
normally run so that the test need only run ffmpeg to compute the
output of vmaf, (or is the preference for every test to generate its
own input?)

Thanks

Kevin
___
ffmpeg-devel mailing list
ffmpeg-devel@ffmpeg.org
http://ffmpeg.org/mailman/listinfo/ffmpeg-devel