Re: [Devel] CFF font

[Werner LEMBERG]

> I've just found another font (embedded subset in the PDF document)
> which produce ERROR 0x6 on FT_New_Memory_Face(...) function.

Are you *sure* that this is a valid font?  Please send me a complete,
small PDF document which contains it.

I believe that this font is not valid.  Reason: The private dict
starts at offset 0x463, having a length of 68 bytes (search for the
sequence `0xCF 0xFA 0xF7 0x12' in font.dat -- 0x12 is the opcode for
the `Private' operator).  But all bytes after offset 0x478 are zero.
The byte before the final zeroes is 0x1E, starting a real number.
Zeroes are valid values in CFF real number representations, so
FreeType continues to parse until the private dictionary size limit
has reached.  Then it correctly emits an error.

BTW, pfaedit (Executable based on sources from 08:28 11-Jan-2004)
immediately segfaults.


    Werner