[pdftex] Incredibly PDF filesize difference by pdflatex compared to ps2pdf

[Reinhard Kotucha]

Hi,
I just looked into the huge file (> 8 MB) produced by dvips, which
Roboco sent me by email.

Dvips did not produce subsets of the fonts.  There are 1524 glyphs in
MinionPro-It and the font looks good.

pdftex and ghostscript produced subsets (112 glyphs).

It seems that the font itself is ok.

That means that there is either a bug in the pdftex subsetting
routine or in any preprocessor.

It's now important to know how the font was loaded into pdftex.  Did
pdftex read the otf file directly or are other programs involved?

Roboco, can you find this out?


The subroutines containing only the token "hhcurveto" are only present
in the output of pdftex.  There is no occurence of the string
"hhcurveto" in the disassembled font extracted from the dvips file. 

BTW., a subroutine like 

dup 149 {
	hhcurveto
	} |

is certainly invalid because the "return" statement is missing.

A program performing font subsetting must be able to understand at
least the subset of PostScript operators used by Type1 fonts.  In
other words, it must provide a minimal PostScript interpreter.

Is it possible that a stack under/overflow within this interpreter
occurs? 

Or is there an array of PostScript keywords and a pointer accidentally
points to "hhcurveto" instead of to "return"?

Regards,
  Reinhard

-- 
----------------------------------------------------------------------------
Reinhard Kotucha			              Phone: +49-511-4592165
Marschnerstr. 25
D-30167 Hannover	                      mailto:reinhard.kotucha at web.de
----------------------------------------------------------------------------
Microsoft isn't the answer. Microsoft is the question, and the answer is NO.
----------------------------------------------------------------------------