On 22/09/16 19:24, Diego Biurrun wrote: > On Tue, Sep 20, 2016 at 10:21:40PM +0100, Mark Thompson wrote: >> On 20/09/16 22:07, Diego Biurrun wrote: >>> On Tue, Sep 20, 2016 at 09:22:06PM +0100, Mark Thompson wrote: >>>> On 20/09/16 19:36, Diego Biurrun wrote: >>>>> ptrdiff_t is the correct type for array strides and similar. >>>>> --- a/libavcodec/svq3.c >>>>> +++ b/libavcodec/svq3.c >>>>> @@ -118,8 +118,8 @@ typedef struct SVQ3Context { >>>>> int mb_x, mb_y; >>>>> int mb_xy; >>>>> int mb_width, mb_height; >>>>> - int mb_stride, mb_num; >>>>> - int b_stride; >>>>> + int mb_num; >>>>> + ptrdiff_t mb_stride, b_stride; >>>> >>>> mb_stride should stay an int. >>> >>> Why? >> >> It's used as: >> >> (a) an offset against the current macroblock for prediction modes. >> >> (b) a multiplier for y offsets to macroblock lines. >> >> (c) a size for macroblock lines. >> >> All of the other things there are ints (well, last one could be size_t), and >> have nothing to do with pointers. > > All of the above sounds stride-ish to me and we seem to use ptrdiff_t > for such variables elsewhere. Ame I wrong?
Probably, I think Mark is the second person finding wrong using ptrdiff there. lu _______________________________________________ libav-devel mailing list libav-devel@libav.org https://lists.libav.org/mailman/listinfo/libav-devel