The file 478-01.ps has shows a difference in the first test between baseline and head (shown with r8013, but the difference started with r8003). The shading varies with resolution, the attached screen shot is a reduced 300 dpi example. The command line used: bin/gs -sDEVICE=ppmraw -sOutputFile=test.ppm -r300 ./478-01.ps
Created attachment 2982 [details] screenshot.png
Created attachment 2983 [details] 478-01_from_CPSI_600_dpi.pdf
Created attachment 2984 [details] 478-01_gs8015_600dpi.png
Our customer that gets frequent snapshots of our code also runs the FTS and CET which includes this file. Bumping priority to correspond to important customer level. I do not see a minor color variation here compared to older (8.57) GS, but the "blue" is entirely missing from the center of the rightmost shaded rectangle in the 478-01 panel ("setsmoothness").
Please read the log message of the revision 8003. The rendering difference is expected. "1 setsmoothness" means that the document allows a 100% error in shading colors. What you observe is a 100% error. Therefore the bug invalid. If a customer needs a special compatibility to another PS interpreter, they should pay for a an instandard feature. Returning the bug to Support.
Created attachment 2987 [details] screen.png If I modify the 478-01.ps file and change the "1 setsmoothness" to "0.5 setsmoothness" strange things happen (see attached screen.png). The command line I used: bin/gs -sDEVICE=ppmraw -sOutputFile=test.ppm -r300 ./478-01.ps
The particular reason for a bad view with "0.5 setsmoothness" is gx_cspace_is_linear_in_line, which checks the lineariry at 2 points only : 0.3 and 0.7 . One can implement checking more points, with the proportional slow down of the rendering, especially with complex color spaces. However I don't think that we want the slowdown. Anyway I'm not clear why should I spend time for this nonsense. I guess the customer didn't request "0.5 setsmoothness". I think it comes from an "academicity" of our Support. Any values less than 1/8 have no practical demand. Well I can code "if (smoothness > 0.125) { check linearity at 0.5} in gx_cspace_is_linear_in_line. Or even if (smoothness < 0.125) smoothness = 0.125; I will change the assignment to Ray for a further resolution.
Another possible resolution for this - it's a dup of Bug 687414.
Regression, but customer can run with limiting setsmoothness to small values.
Change to ASSIGNED. I will re-test.
Tested with HEAD and this looks OK (matching Adobe).