A few files threw rangecheck and continues with pdfwrite and have some graphics missing, compared to the silent ppmraw results. tests_private/pdf/PDF_1.7_FTS/fts_17_1700.pdf.pdf tests_private/pdf/PDF_1.7_FTS/fts_17_1701.pdf.pdf tests_private/pdf/PDF_1.7_FTS/fts_17_1714.pdf.pdf tests_private/pdf/PDF_1.7_FTS/fts_17_1715.pdf.pdf tests_private/pdf/PDF_1.7_FTS/fts_18_1805.pdf.pdf tests_private/pdf/PDF_1.7_FTS/fts_18_1812.pdf.pdf That probably means there is room for improvement in the pdfwrite code, and that the current pdfwrite output for these files aren't correct. I came across this when I accidentally made the rangecheck into a segfault in one of my tests on earlier versions of the j2k code. It is possible that this may already been reported by Marcos?
At least the first of these, almost certainly all of them, are a duplicate of Bug #694371 "pdfwrite doesn't handle 16 bpc images". In this particular case the 'images' are actually soft masked images, and that means we can't fallback either. 16 bpc images are rare so I'mleaving teh original enhancement request in place, if we start to see more 16 bpc images, or this becomes a customer problem, we'll revisit it. *** This bug has been marked as a duplicate of bug 694371 ***