Adding "/Type /Info" as an info dictionary entry produces an entry in the info
dict whose value is a name(pdf data type). This is invalid (as names are not
allows as values within the info dict), and causes some PDF Libraries the throw
errors when reading the information. 

In my case i have an application using the Adobe PDF Library 7.0 that throws an
error("Could not create internal representation of XMP metadata") when calling
PDDocGetXAPMetadataProperty. When i remove the invalid entry from the info dict,
everything works properly.

I did look in the svn repository and it appears that the Trunk code should still
have the same issue.

The following is what i updated in 0.20.5 to work around the issue

From org.apache.fop.pdf.PDFInfo

    public byte[] toPDF() {
        String p = this.number + " " + this.generation
                   + " obj\n<< /Type /Info\n/Producer (" + this.producer
                   + ") >>\nendobj\n";
        try {
            return p.getBytes(PDFDocument.ENCODING);
        } catch (UnsupportedEncodingException ue) {
            return p.getBytes();
        }       
    }

should be changed to(just removing the "/Type /Info")
    public byte[] toPDF() {
        String p = this.number + " " + this.generation
                   + " obj\n<< \n/Producer (" + this.producer
                   + ") >>\nendobj\n";
        try {
            return p.getBytes(PDFDocument.ENCODING);
        } catch (UnsupportedEncodingException ue) {
            return p.getBytes();
        }       
    }

Thanks,
john