[gs-devel] ESP Ghostscript merge

[Ralph Giles]

Till,

Thanks for maintaining the merge branch these past few weeks. I've now 
looked at the remaining differences we were seeing in the cluster 
regression. The following files were showing differences 
with ppmraw at 600 dpi:

  Bug688467.ps
  Bug688308.ps
  bc_02a01_cal.pdf
  besttest.pdf

The differences are all the same effect: a one-line shift in a scaled 
image. I consider then visually equivalent. Also, with your merge
from trunk r8024, the differences went away, and the regression shows
the two runs as producing bit-identical output.

I conclude therefore that the differences are an indeterminism, perhaps 
in the image scaling or DCT decode. In light of this, it's ok to proceed.

Please merge your patches from the branch down into trunk.

Cheers,
 -r