We are creating a PostScript file and using ghostscript 8.50 to
distill it to PDF.  The document uses an Adobe OpenType font,
PostScript (CFF) flavor, non-CID.  Our software extracts the CFF,
wraps it as a FontSet resource as shown in Adobe Technical Note #5176
(CFF spec), and embeds it in the PS file.  We access characters by
CMap.

When ghostscript distills it, it turns the font into a Type 3 font (at least,
that's what Adobe Reader reports), which then looks lousy in Adobe Reader, as
noted in several other places in this forum.  (Specifically, we see irregular
baselines that vary with magnification in the Reader.)  Same results in Foxit
Reader and gsview.

Is this a ghostscript bug?  Our failure to use the right switches?  Is there
anything we can do to change ghostscript's behavior?  I can understand why it
might mess with TrueType-flavored OTF data (which we present as Type 42), but
why should it change the PostScript flavor?

Here are the ghostscript switches we are using to create the PDF:

    -q -dNOPAUSE -dBATCH -sDEVICE=pdfwrite -sOutputFile=xxx.pdf \
    -c .setpdfwrite -f input.ps

By default, this creates PDF version 1.3.  (I broke the command into
two lines for illustration in this forum only.)

When we use Acrobat Distiller, the output is fine.

We have also experimented with another switch: -dNOCACHE.  This produces
better-looking PDF, and "only" makes the PDF file **20 times as
big**!  Think what that would do to your college guide, etc.!

We run our software on Windows, Unix and Linux.

Thanks for any help you can provide.

Mark DeVries
XyEnterprise Inc.