Re: Printing problem (Conflicting font name?)

["Berthold K.P. Horn" <bkph@ai.mit.edu>]

At 08:35 AM 98/10/15 +0200, you wrote:

>Concernant « Re: Printing problem (Conflicting font name?) », Berthold Horn écrit : «

>» As I stated before, there is another state that is preserved across jobs:
>» the glyph cache, which is organized on FontName, FontMatrix and
>» UniqueID.  Which is why it is a bad (TM) idea to call a modified font
>» by the same PS FontName as the original.  Some DVI drivers do this...

>Are you saying that using a subsetted font without changing its name
>could suppress some glyphs for further jobs (like in acrobat)?

>That sounds terrific!

Yes, a well known bug in PS produced by DVIPS.  Which is why
you should either (i) not include fonts (tell DVIPS they are printer
resident in psfonts.map) and let Distiller find them, or (ii) at least
suppress partial font downloading (-j0).  Another solution would be
for DVIPS to rename the font with a random prefix (the way
Distiller does when it subsets fonts), but this also requires addition
of some new fiellds (like /BaseFont) to the PS header of the font.

Theoretically reencoding a font should not require renaming it,
by the way, since the glyph cache is supposed to be organized
on (i) UniqueID, (ii) FontName, (iii) FontMatrix and (iv) glyph name
(*not* character code), but I think it is safer to use a different
name even in this case (e.g. ATM's glyph cache ignores UniqueID!).

Regards, Berthold.

Berthold K.P. Horn
Cambridge, MA		mailto:bkph@ai.mit.edu