[texhax] PDF produced by pdflatex rejected by Lulu.com

[Reinhard Kotucha]

On 2012-02-09 at 20:53:17 -0800, Donald Arseneau wrote:

 > "Niall Mansfield" <texhax at uit.co.uk> writes:
 > 
 >> We produced a book in PDF using pdflatex, and sent it to
 >> Lulu.com to print a hardcopy proof.
 >> 
 >> Days and days later (grrrrrrrrr) we got a message saying
 >> they had a problem with the file and were refunding our
 >> payment.  The only information they provided was
 >> 	"RIP error"
 >> and
 >> 	"It looks like there's an issue with the fonts in
 >> 	your PDF."
 >> 
 >> pdffonts says (below) that all fonts are embedded, so I
 >> can't think what's wrong.
 >> 
 >> Has anybody either:
 >> a. successfully printed LaTeX documents with Lulu?  or ...
 >> b. had similar problems?  Solutions?
 >> 
 >> 	Thanks,
 >> 	Niall Mansfield
 >> 
 >> name                                 type              emb sub uni object ID
 >> ------------------------------------ ----------------- --- --- --- ---------
 >> NJKBOS+URWPalladioL-Roma             Type 1            yes yes no       3  0
 >> ZIBQYN+URWPalladioL-Bold             Type 1            yes yes no       8  0
 >> EEHSNB+URWPalladioL-Ital             Type 1            yes yes no      19  0
 >> TZDPKW+NimbusSanL-Bold               Type 1            yes yes no      23  0
 >> YMPRVZ+NimbusMonL-Regu               Type 1            yes yes no      36  0
 >> AHFKWD+NimbusSanL-ReguCond           Type 1            yes yes no      52  0
 >> OWTLWS+URWPalladioL-BoldItal         Type 1            yes yes no     106  0
 >> PalatinoLinotypeItalic               CID TrueType      yes no  yes    127  0
 >> GillSans                             CID TrueType      yes no  yes    132  0
 >> PalatinoLinotypeItalic               CID TrueType      yes no  yes    143  0
 >> GillSans                             CID TrueType      yes no  yes    148  0
 > 
 > 
 > Could the problem be the duplicate PalatinoLinotypeItalic and GillSans?

BTW, these duplicate entries are harmless but can confuse some
programs though.  I encountered such problems when running pdfsizeopt
on the VnTeX font sample files.  Further investigation showed that
if pdffonts lists different objects for the same font, these objects
don't contain the font file itself but references to a font file and
an encoding vector.

This happens, for instance, if a font file contains Roman and Small
Caps and both are used in a document.  pdfTeX inserts each font only
once but needs different encodings in order to access all glyphs.
This is correct behavior and compliant with the PDF specification.

Obviously pdffonts doesn't follow the link to the actual font.  It
would then notice that both links point to the same font-file object.

As far as pdfsizeopt is concerned, what I don't understand is why it
rejects only some of such files complaining about duplicate fonts,
while its author, Peter Szabó, could compile all files from

  http://vntex.sf.net/fonts/samples

seamlessly on his laptop at EuroTeX 2009 [1].  

Anyway, all other programs (PDF viewers and to-PostScript converters)
I'm aware of, even xpdf itself (from which pdffonts is derived), find
the actual fonts and produce proper output.

And regarding Lulu.com, I'm wondering how one can bother users with
such a stupid and useless error message:

 >>  "It looks like there's an issue with the fonts in your PDF."

It's foreseeable that the hotline runs hot.  Looks like a self made
denial-of-service attack.  Do they still respond to mails?  I can't
imagine.  I'm also wondering how a machine which is only aware of ones
and zeroes can provide an error message containing "It looks like
there's an issue with ...".  Is it that what people call "fuzzy logic"
or is it a first attempt to support what's called "artificial
intelligence"?  Or do they simply use trinary logic, supporting the
states "true", "false", and "dontknow"?

[1] http://pdfsizeopt.googlecode.com/files/pts_pdfsizeopt2009.psom.pdf

Regards,
  Reinhard

-- 
----------------------------------------------------------------------------
Reinhard Kotucha                                      Phone: +49-511-3373112
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.
----------------------------------------------------------------------------