[FFmpeg-devel] [PATCH 24/24] avcodec/mpegvideo: Don't use ScanTable where unnecessary

[Andreas Rheinhardt]

Michael Niedermayer:
> On Thu, Oct 20, 2022 at 07:24:36PM +0200, Andreas Rheinhardt wrote:
>> For the intra_[hv]_scantables, only ScanTable.permutated
>> is used, so one only needs to keep that.
> [...]
>> diff --git a/libavcodec/mpeg4videoenc.c b/libavcodec/mpeg4videoenc.c
>> index 8e6e35b927..bf499a2206 100644
>> --- a/libavcodec/mpeg4videoenc.c
>> +++ b/libavcodec/mpeg4videoenc.c
>> @@ -175,7 +175,7 @@ static inline int decide_ac_pred(MpegEncContext *s, int16_t block[6][64],
>>                      ac_val1[i + 8] = level;
>>                  }
>>              }
>> -            st[n] = s->intra_h_scantable.permutated;
>> +            st[n] = s->intra_h_scantable;
>>          } else {
>>              const int xy = s->mb_x - 1 + s->mb_y * s->mb_stride;
>>              /* left prediction */
>> @@ -197,7 +197,7 @@ static inline int decide_ac_pred(MpegEncContext *s, int16_t block[6][64],
>>                      ac_val1[i + 8] = block[n][s->idsp.idct_permutation[i]];
>>                  }
>>              }
>> -            st[n] = s->intra_v_scantable.permutated;
>> +            st[n] = s->intra_v_scantable;
> 
> Iam thinking that replacing 
> s->intra_v_scantable.permutated
> by
> intra_v_scantable
> is semantically feeling inferrior to calling it
> permutated_intra_v_scantable
> or something like that
> 
> The same probably applies to te other patches too
> iam not happy about the long name but having consistent
> naming between scantables (as in the spec) and scantables permutated
> for idct optimizations could improve readability of the code
> 

Ok, will send a v2 with this changed.

- Andreas