[XviD-devel] possible bug :(

[Christoph Lampert]

On Wed, 31 Jul 2002, Radoslaw 'sysKin' Czyz wrote:

> Hi
> 
> > So B-frames are still bugfree ;-) unless somebody comes up with
> > something new...
> 
> Sorry, it seems I did. But it's not very important, please don't reset
> the counter ;) (even if that's cheating).
> Of course I may be wrong.
> 
> In direct mode with inter4v (motion_est.c line 3213) there is:
>   sad8bi(frame->image.y + 2*(i+(k&1))*8 + 2*(j+(k>>1))*8*edged_width,
>          get_ref_mv(f_ref->y, f_refH->y, f_refV->y, f_refHV->y,
>                     2*(i+(k&1)), 2*(j+(k>>1)), 8, &mb->mvs[k], edged_width),
>          get_ref_mv(b_ref->y, b_refH->y, b_refV->y, b_refHV->y,
>                     2*(i+(k&1)), 2*(j+(k>>1)), 8, &mb->b_mvs[k], edged_width),
>                                                         edged_width);
> 
> where is:
>   2*(i+(k&1)) , shouldn't be
>   (2*i+(k&1)) ?
> 
> The same for 2*(i+(k>>1)) . It happens in both get_ref_mv and in the
> first parameter of sad8 (6 times total)

You are absolutely right. And although it might not cause a crash 
it's still very bad evil nasty ugly bug, simply because it's there. 
So I just commited the fix. 

gruel


P.S. Now this doesn't look too impressive anymore... 
*******************************************************************
*                   B-frame bugfree counter                       *
*                   =======================                       *
*                                                                 *
*    No bug in B-frames found for   /-----------\   day(s).       *
*                                   |   #####   |                 *
*                                   |  #######  |                 *
*                                   |  ##   ##  |                 *
*                                   |  ##   ##  |                 *
*                                   |  #######  |                 *
*                                   |   #####   |                 *
*                                   \-----------/                 *
*******************************************************************

-- 
Christoph H. Lampert chl@math,uni-bonn,de | Diese Signature wurde maschi-     
Beringstr. 6, Raum 14 Tel. (0228) 73-2948 | nell erstellt und bedarf
Sprechstunden: keine, aber meistens da    | keiner Unterschrift. AZ 27B-6