[NTG-context] Problems with Palatino sans serif

Taco Hoekwater taco at elvenkind.com
Mon Oct 20 14:45:50 CEST 2008

Mojca Miklavec wrote:
>> For this, can you send me a private message with before & after set
>> of pdfs?
> http://www.mars2008.si/taco/
>> 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
   /FontDescriptor 344 0 R
   /LastChar 246
   /Widths 343 0 R
   /FirstChar 2
   /ToUnicode 340 0 R
   /Encoding 342 0 R
342 0 obj
   [46/period 53/five 71/G 80/P 84/T 97/a 101/e
   104/h/i/j/k/l/m/n/o 114/r/s/t 121/y]
344 0 obj
<</StemV 84
   /FontFile 339 0 R
   /Flags 4
   /Descent -282
   /FontBBox[-166 -283 1021 943]
   /Ascent 715/CapHeight 680/XHeight 469
   /ItalicAngle -9

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


at the start of /Differences, then suddenly the slanted accented z-s

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,

More information about the ntg-context mailing list