[pdftex] Fixing Encoding object problems [was: AR 5 not working with pdftex?]

Tom Kacvinsky tjk at ams.org
Sat Jun 9 19:03:05 CEST 2001

Hi Hans,

> i wonder, if pdftex has a dangling pointer, we could add the vector at the
> end of the file with that number as object and cross our fingers that
> acrobat would 'reconstruct' the file, but i must admit that i never saw
> acrobat being able to really correct / reconstruct a corrupted file.

I see that the object that is supposed to have the encoding vector is marked
as free in the xref table. So, one could add the encoding object to the end
of the PDF file (before the xref table), and fix the both the xref table and
startxref value, simply because even though Acrobat will "fix" the PDF file,
when it is saved, one gets all of the extra stuff Acrbat likes to insert.

> Also, say that we should copy the font descriptor, add an encoding object
> [both ascii], add an additional ref table, then we would have an updated
> file [second obj ref number being 1 instead of 0] and this would be valid.
> Of course the problem would be to identify the correct vector. A bunch of
> such files could then be optimized by acrobat.

The Font object already has "/Encoding # 0 R" in it, it is just that # is
marked as free in the xref table.  So we don't have to add an additional
xref table and mark the new objects as having generation 1 (instead of 0).

See above for why.


More information about the pdftex mailing list