Mojca Miklavec wrote:
For this, can you send me a private message with before & after set of pdfs?
If possible, ones generated with \nopdfcompression (because that is easier to debug). Afaict, it is unrelated to this problem, but still definately a bug.
The file is 26MB now. I will try to see what I can do to cut it down to some smaller size & still showing the bug, before filling the disk quota :) :) :) by sending uncompressed PDF.
I fetched the files (so you can delete them) and I am tempted to believe this is an AR (distiller) bug, because the /Encoding in the generated PDF is incomplete, and it was probably correct in the z-before.pdf (I did not decompress the file, but it must have been ok, otherwise that file would have displayed incorrectly as well). The "Slanted" entry in the z-after.pdf has: 341 0 obj <
endobj 342 0 obj <
endobj 344 0 obj <
endobj
Notice the big difference between the /Differences and the /CharSet. Despite the name of /Differences, this normally lists all the glyphs in the subset. When I manually add 14/Zcaron/zcaron at the start of /Differences, then suddenly the slanted accented z-s reappear. I suspect Acrobat Pro gets confused about the encoding because it probably has to merge more than one subset into a single one. Perhaps it believes all encodings are equal. This you could test by looking at an uncompressed generated pdf's Font objects in a text editor. Best wishes, Taco