[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
<</Subtype/Type1
   /FontDescriptor 344 0 R
   /LastChar 246
   /Widths 343 0 R
   /BaseFont/HFKCGX+URWPalladioL-Roma-Slant_167
   /FirstChar 2
   /ToUnicode 340 0 R
   /Encoding 342 0 R
   /Type/Font
 >>
endobj
342 0 obj
<</Differences
   [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]
   /BaseEncoding/WinAnsiEncoding
   /Type/Encoding
 >>
endobj
344 0 obj
<</StemV 84
   /FontName/HFKCGX+URWPalladioL-Roma-Slant_167
   /FontFile 339 0 R
   /Flags 4
   /Descent -282
   /FontBBox[-166 -283 1021 943]
   /Ascent 715/CapHeight 680/XHeight 469
   /ItalicAngle -9
   /CharSet(/A/B/C/D/E/F/G/H/I/J/K/L/M/N/O/P/R/S/Scaron/T/U/V/W/Z\
           /Zcaron/a/b/bracketleft/bracketright/c/caron/colon/comma\
           /d/e/eacute/eight/endash/exclam/f/fi/
           /five/four/g/guillemotleft\
           /guillemotright/h/hyphen/i/j/k/l/m/n/nine/o/odieresis/one\
           /parenleft/parenright/period/question/r/s/scaron/seven/six\
           /t/three/two/u/v/y/z/zcaron/zero)
   /Type/FontDescriptor
 >>
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






More information about the ntg-context mailing list