[luatex] CIDSet in PDF/A documents

Reinhard Kotucha reinhard.kotucha at gmx.de
Tue Jun 23 03:25:59 CEST 2015



Hans wrote

 > pdf/a demands a cidset but we will forget about this till we find a
 > proper example
 > there is still some reported problem with some stream objects not
 > properly being formatted cf. pdf/a but it is not clear what is
 > going on there

This message is now five years old.  Any new perceptions?

I'm using a TrueType font (CharisSIL) and one of five PDF/A validators
complains about a bad CIDSet.

I created a small PDF file which only contains the string "abc",
extracted the TTF from the PDF file, and disassembled it with TTX.
Then I assembled the CIDSet manually according to the instructions
given in the PDF/A-1 specification (ISO 19005-1).  I've got the same
result as LuaTeX, hence it's unclear to me what's going wrong.

The PDFtron


validator sais

  <Error Code="e_PDFA356" Message="CIDSet in subset font is incomplete" Refs="95, 101"/>

I tend to believe that the validator is wrong.  On the other hand
PDFtron offers software which creates PDF/A files and I can't imagine
that their validator complains about their own products.

Did anybody investigate?  The nasty thing is that PDF/A is for long
term preservation and any file we create today has to comply with the
standard unconditionally.  And for us TeX users, the fact that there
are zillions of invalid PDF/A files around just because old versions
of the Acrobat preflight tool ignored most errors, is not an excuse.
We should do better.

>From the results of my own investigations I deduce that LuaTeX
provides a standard compliant CIDSet.  Maybe different people
interpret the standard in a different way.  But it would be nice to
know whether somebody else investigated this issue already.


Reinhard Kotucha                            Phone: +49-511-3373112
Marschnerstr. 25
D-30167 Hannover                    mailto:reinhard.kotucha at web.de

More information about the luatex mailing list