[FFmpeg-devel] Various small fixes

Måns Rullgård mans
Sat Dec 13 16:36:01 CET 2008


Uoti Urpala <uoti.urpala at pp1.inet.fi> writes:

> On Sat, 2008-12-13 at 14:52 +0000, M?ns Rullg?rd wrote:
>> "Anders Gr?nberg" <galileo.m2 at gmail.com> writes:
>> > +++ libavcodec/imgconvert.c	(working copy)
>> > @@ -783,7 +783,7 @@
>> >      dst_pix_fmt = -1;
>> >      min_dist = 0x7fffffff;
>> >      for(i = 0;i < PIX_FMT_NB; i++) {
>> > -        if (pix_fmt_mask & (1 << i)) {
>> > +        if (pix_fmt_mask & (1ULL << i)) {
>> >              loss = avcodec_get_pix_fmt_loss(i, src_pix_fmt, has_alpha) & loss_mask;
>> >              if (loss == 0) {
>> >                  dist = avg_bits_per_pixel(i);
>> 
>> This is madness.  What do you believe the benefit of that to be?
>
> pix_fmt_mask is int64_t.

I should have checked that.  I assumed it wasn't.

> If int is 32-bit then (pix_fmt_mask & (1 << i)) can access at most
> 32 bits of it.

Of course.

I still blame the patch submitter for not explaining a single line.

-- 
M?ns Rullg?rd
mans at mansr.com




More information about the ffmpeg-devel mailing list