[pdftex] inclusion of PDF file fails

[Thanh Han The]

On Mon, Jan 23, 2012 at 8:31 AM, Werner LEMBERG <wl at gnu.org> wrote:
>>  > > While the `CenturySchL-Roma' font embedded in `BFEx.pdf' is
>>  > > version 1.06 (file `c059013l.pfb' from ghostscript), pdftex
>>  > > uses `uncr8a.pfb' from TeXLive, which is version 1.05.
>>  > > Obviously, version 1.05 misses all Cyrillic glyphs.  IMHO, the
>>  > > right solution is to allow font merging only if the font
>>  > > versions are identical.
>>  >
>>  > Is someone going to fix this?
>>
>> PdfTeX already merges fonts only if they have the same /FontName and
>> /UniqueID.
>
> But the concept of /UniqueID has been abandonded by Adobe a very long
> time ago:
>
>  http://typophile.com/node/18749
>
> (This link is from 2006.)  So it's actually another bug if pdftex
> relies on it.
>
>> However, ghostschript provides fonts with are buggy in this respect.
>> They contain additional glyphs but have the same /FontName and
>> /UniqueID as the original URW fonts.  I don't know how pdfTeX should
>> recognize this.
>
> As I've already mentioned: Use the font's version (1.05 vs. 1.06 in
> the example given by me)!  If someone produces fonts which are
> different but have exactly the same identifier, nothing can be done
> except punishing the font developer.

pdftex considers two fonts are the same if they have the same
/FontName (/UniqueID is not used).

comparing font version is certain a better way to do it, but the
change is not something like a few lines of code. pdftex must parse
the font file (from disk) and the font embedded in the pdf. Both are
doable but not trivial due to the way pdftex handles fonts now.

In the meantime, perhaps setting \pdfinclusioncopyfonts=1 (as the
default value at engine level) is the safe thing to do (with the risk
that some people will blame pdftex for increasing pdf output size).

Thanh