[tex-k] Ongoing problem with use of Small-caps, OSF etc.

[Pierre MacKay]

I call your attention to what happens if you use small caps, old style 
figures, or ff ligatures in a PDF produced with ghostscript (86.3).

When such a PDF is displayed in Acrobat 8 or 9, these characters simply 
disappear and seem to be replaced by the MissingWidth spacing which is 
provided in Ghostscript-generated subset fonts.  My guess would be that 
every character from the Adobe Expert character set is treated the same 
way.  The attached file, with extracted subsets of small-caps characters 
from Monotype Times New Roman CS OSF shows how very differently the 
Adobe.com distiller and the Ghostscript pdf module handle Adobe Expert 
character set glyph names.  Adobe clearly recodes them in some way, 
though it would be difficult to understand how or why.  A few expressly 
identified small caps are permitted in Unicode, in page 1D00, but by no 
means a full set. ff ligatures are provided in page FB00.  Old style 
figures are provided for nowhere, so /"ToUnicode" cannot be providing 
canonical Unicode values for most of what is missing.

One solution would be to go into the CS and OSF files and change the 
character names to vanilla ASCII, so that they would not show up to the 
distiller as peculiar.  Can anyone think of a less painful approach?  
Could the Ghostscript people be persuaded to find some approach similar 
to what Adobe.com's distiller provides?

There is also the problem that archives of PDFs from former years are 
now rendered useless, because any cure we can find will not work for 
these earlier files unless they are all reset and redistilled.

Pierre MacKay


Acrobat Reader 7 and 6 are not afflicted by this suppression of CS and 
OSF.  The ideal would be to persuade Adobe to provide some sort of 
exception that did not require recoding these characters, but they 
assured me that they would do just that after I filed a bug report over 
Reader 8, and clearly nothing has been done.

I had thought that the problem might have been caused by my use of the 
Monotype TimesNewRomanSuperFont, with its huge encoding vector 
(something like 400 C-1 characters beyond the first 255) but the font I 
am now using as the source for CS and OSF is a normal 255-character 
Type1 font.  It is the character names that are causing the problem.
I still think people should probably be advised NOT to use Monotype 
SuperFonts.

Pierre MacKay

-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: font-handling.txt
URL: <http://tug.org/pipermail/tex-k/attachments/20091003/a858aec1/attachment.txt>