The customer reports: Below is a "diff" of a fix that prevents an infinite loop (which i saw trying to render with Interpolation a PDF that had a 2x1-pixel image). There's also another fix found by observation. I haven't seen any consequence from the second bug yet. You're welcome.
Created attachment 5776 [details] patch
Thank you for the patch. Could you supply the file with the 2x1 image? I agree that we should not do the interpolation process if iss.WidthOut == iss.WidthIn && iss.HeightOut == iss.HeightIn . I don't agree though with iss.WidthIn < 2 || iss.HeightIn < 2. If I have an Nx1 or 1xN ( column or row ) image, then interpolation will be different than pixel replication/removal. We could have this be (iss.WidthIn < 2 && iss.HeightIn < 2), but then your 2x1 image would have its issue. I would like to understand why that occurs rather than mask it. The other issue was fixed in rev 9873.
Created attachment 5942 [details] smallimage.ps
Assigning to Dave to investigate infinite loop issue with small image.
Unable to reproduce on Intel PC, Intel Mac, or PowerPC Mac with HEAD, either with smallimage.ps (contains a 3x8 image) or with a self-made PDF with a 2x1 image. Testing included setting a breakpoint at the code related to the patch and verifying the breakpoint was reached.
I'm not able to reproduce this with the current head (and without the supplied patch). I'm closing the bug but will reopen if I can find a command line that fails.