Hi,

On Mon, Sep 4, 2017 at 10:11 PM, James Almer <jamr...@gmail.com> wrote:

> On 9/4/2017 11:00 PM, Ronald S. Bultje wrote:
> > Hi,
> >
> > On Mon, Sep 4, 2017 at 12:11 PM, Michael Niedermayer
> <mich...@niedermayer.cc
> >> wrote:
> >
> >>          if(type){
> >> +            int ld, cbd, crd;
> >>              pred_mv(s, &mx, &my, 0, left, top, tr);
> >> -            l += get_symbol(&s->c, &s->block_state[32], 1);
> >> +            ld = get_symbol(&s->c, &s->block_state[32], 1);
> >> +            if (ld < -255 || ld > 255) {
> >> +                av_log(s->avctx, AV_LOG_ERROR, "Invalid (Out of range)
> >> intra luma block DC difference %d\n", ld);
> >> +                return AVERROR_INVALIDDATA;
> >> +            }
> >> +            l += ld;
> >>              if (s->nb_planes > 2) {
> >> -                cb+= get_symbol(&s->c, &s->block_state[64], 1);
> >> -                cr+= get_symbol(&s->c, &s->block_state[96], 1);
> >> +                cbd = get_symbol(&s->c, &s->block_state[64], 1);
> >> +                crd = get_symbol(&s->c, &s->block_state[96], 1);
> >> +                if (cbd < -255 || cbd > 255 || crd < -255 || crd >
> 255) {
> >> +                    av_log(s->avctx, AV_LOG_ERROR, "Invalid (Out of
> >> range) intra chroma block DC difference %d, %d\n", cbd, crd);
> >> +                    return AVERROR_INVALIDDATA;
> >> +                }
> >> +                cb += cbd;
> >> +                cr += crd;
> >>              }
> >
> >
> > I recognize the great improvements in your messages. They are much better
> > than before. They are still not appropriate for display to end users.
> > Please use ff_tlog(), as was suggested in the original thread.
>
> Now that i looked at ff_tlog() closely it's only enabled if -DTRACE is
> used during compilation, and then it prints stuff at the trace log
> level, which is even lower than debug and used to print a bunch of
> assorted values that are not error log messages.


You're right, ff_dlog() seems better (see libavcodec/internal.h).

I also agree with y'all, let's stop discussing this further, we're all
working on real, actually important stuff, this particular issue really
isn't worth our time...

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

Reply via email to