[FFmpeg-devel] [PATCH] avfilter/vf_transpose: Fix regression with packed pixel formats

[James Almer]

On 1/27/2018 10:50 PM, Michael Niedermayer wrote:
> Regression since: c6939f65a116b1ffed345d29d8621ee4ffb32235
> Found-by: Paul B Mahol <onemda at gmail.com>
> Signed-off-by: Michael Niedermayer <michael at niedermayer.cc>
> ---
>  libavfilter/vf_transpose.c | 2 +-
>  1 file changed, 1 insertion(+), 1 deletion(-)
> 
> diff --git a/libavfilter/vf_transpose.c b/libavfilter/vf_transpose.c
> index 1e1a5c4b89..bf2ab7eb18 100644
> --- a/libavfilter/vf_transpose.c
> +++ b/libavfilter/vf_transpose.c
> @@ -217,7 +217,7 @@ static int config_props_output(AVFilterLink *outlink)
>  
>      s->hsub = desc_in->log2_chroma_w;
>      s->vsub = desc_in->log2_chroma_h;
> -    s->planes = desc_in->nb_components;
> +    s->planes = (desc_in->flags & AV_PIX_FMT_FLAG_PLANAR) ? desc_in->nb_components : 1;
>  
>      av_assert0(desc_in->nb_components == desc_out->nb_components);

If there are currently no tests for this filter (or at least not for
packet formats), then one should be added.

Lavfi's current fate coverage is unacceptably low. Something like two
third of the codebase remains untested according to LCOV.
At this point no software filter should be added without a fate test for
at least a basic usage example using either the fate generated synthetic
sources or any samples in the suite. Same with new features for new filters.