ReconBuffer Test Issue

[GoogleCodeExporter]

Failure 1

Im seeing an issue with the reconbuffer test where it will fail for two pass 
best quality for a target bit rate of 128 for src16 for Win 32/64, Lin 32/64 
and IMac 64. The first instance the fail seems to occur for is:

Change I30511fc8: Badly placed initialization of rolling rate monitors.
https://review.webmproject.org/#change,604

The only difference between this and the last commit seems to be in onyx_if.c.  
Reverting onyx_if.c back to its previous state will fix the failure.

Failure 2

Seems to happen for two pass good and two pass best quality for src16 at a 
target bit rate of 128 for Win 32/64, Lin 32/64 and IMac 32/64

The first instance the fail seems to occur for is:

Change I6b2583a8: enable trellis quantization for 2nd order blocks
https://review.webmproject.org/#change,655

The only difference between this and the last commit seems to be in encodemb.c. 
 Reverting encodemb.c back to its previous state will fix the failure.

Copies of the Test and source code being used can be found here:
Windows: \\filer\home\jamesberry\potential-bugs\2010-10-05-ReconBuffer
Linux  : ~jamesberry/potential-bugs\2010-10-05-ReconBuffer
Mac    : /home/jamesberry/potential-bugs\2010-10-05-ReconBuffer 

Original issue reported on code.google.com by jamesbe...@google.com on 5 Oct 2010 at 8:56

[GoogleCodeExporter]

Issue verified and resolved by 
Change I5cce5da0: Centralize mb skip state calculation
Change-Id: I5cce5da0dbc2d22f7d79ee48149f01e868a64802

-Closed

Original comment by jamesbe...@google.com on 12 Oct 2010 at 1:56

[GoogleCodeExporter]

Original comment by jamesbe...@google.com on 18 Apr 2011 at 2:18

• Changed state: Fixed

[GoogleCodeExporter]

Original comment by albe...@google.com on 8 Mar 2012 at 12:10

• Added labels: Type-Bug
• Removed labels: Type-Defect