The customer reports and I've verified that Ghostscript renders the attached PDF file with damaged images. This is a regression, gs8.54 renders the file correctly but gs8.70 and gs head do not. The exact revision that broke this is hard to narrow down: 8527: works 8528: does not build 8529: does not build 8530: does not build 8531: locks up ... 8542: locks up 8543: broken images The command line I'm using for testing: bin/gs -sDEVICE=ppmraw -o test.ppm ./SB_A0_x1_arctic_700.pdf
Created attachment 5515 [details] SB_A0_x1_arctic_700.pdf
The suspected changes is: r8539 | leonardo | 2008-02-22 12:18:08 -0800 (Fri, 22 Feb 2008) | 32 lines Fix (images) : Improve coordinate precision when scaling an image (continued 6). Of course, this is part 6 of the changes, so it might not be directly related to this change. Since these changes are primarily in the image interpolation, I tried with -dNOINTERPOLATE and the image is correct. This is a work around until the bug can be fixed.
With wts devices removed and patch for the rev. 8543 back-ported I was able to build revs. 8528, 8529 and run the sample file. The former works OK but the latter has broken images. So it's rev. 8529 .
Created attachment 5558 [details] dogs-i4.pdf -- simplified sample file This is a simplified sample file that stretches an an interpolated image to the whole page and places red background under the image. At 300 dpi Ghostscript renders it as a red page with 2 image stripes at the bottom of the page.
This is a problem in the y positioning of interpolated images when the image is bottom up -- an extra inversion of a sign moving the image down past the bottom of the band (for all but the bottom band). Sometimes the image would be partly visible, shifted down. I am testing a patch for this now.
Fixed rev 10542. This is a simple patch and can easily be applied to older versions of gxiscale.c
Changing customer bugs that have been resolved more than a year ago to closed.