[FFmpeg-cvslog] r18852 - trunk/libavcodec/msrledec.c

[Mike Melanson]

Reimar D?ffinger wrote:
> On Sat, May 16, 2009 at 04:58:51PM +0300, Kostya wrote:
>> On Sat, May 16, 2009 at 03:40:21PM +0200, Reimar D?ffinger wrote:
>>> On Sat, May 16, 2009 at 01:35:40PM +0300, Kostya wrote:
>>>> On Sat, May 16, 2009 at 10:37:01AM +0200, Reimar D?ffinger wrote:
>>>>> On Sat, May 16, 2009 at 08:39:08AM +0200, kostya wrote:
>>>>>> Author: kostya
>>>>>> Date: Sat May 16 08:39:08 2009
>>>>>> New Revision: 18852
>>>>>>
>>>>>> Log:
>>>>>> Fix off-by-one error in MS RLE decoder which may result into writing past
>>>>>> picture buffer.
>>>>> FATE doesn't like it, can you confirm the FATE sample still decodes
>>>>> correctly?
>>>> Which one is that? If that is clock.avi then you'll see that clock is
>>>> centered now. And if you take workcycl-64color.avi you'll see that black
>>>> line on the top has gone now.
>>> clock.avi is one of those, but it is also
>>> http://samples.mplayerhq.hu/fate-suite/aasc/AASC-1.5MB.AVI
>>> (based on http://samples.mplayerhq.hu/V-codecs/AASC/AASC.AVI)
>>> http://samples.mplayerhq.hu/V-codecs/tscc/oneminute.avi
>>> http://samples.mplayerhq.hu/fate-suite/tscc/2004-12-17-uebung9-partial.avi
>>> (based on http://samples.mplayerhq.hu/V-codecs/tscc/2004-12-17-uebung9.avi)
>>> all of which changed FATE test results.
>> Of course - they all use the same decompressor.
> 
> Which doesn't answer the question: does the code need to be fixed or
> should Mike update the checksums because it only now decodes correctly?

I visually verified that all 4 tests still look good. So I updated each.

-- 
     -Mike Melanson