For those, who want to test PDF viewer compatibility I have a demo PDF which demonstrates Type 1 "step 2" implemented with solution 2 (multiple descendant fonts with dynamic encoding build-up).
http://people.apache.org/~jeremias/fop/type1-demo/ - [1] font-type1-demo-before.pdf (revision 627678, before I added the AFM stuff, i.e. step 1) - [2] font-type1-demo-step1.pdf (current FOP Trunk HEAD) - [3] font-type1-demo-step2.pdf (my local working copy) The PDFs are not compressed so it's easy to see what's happening inside and what the differences are. The evolution is clearly visible. With [2] the Baskerville Cyrillic fonts starts to display cyrillic characters from its primary encoding. With [3] you get a few more characters with Baskerville. The biggest leap forward is with the URW Gothic L font which contains characters from various languages. Now I have to do the same adjustments like for PDF for PostScript also. Please note that there's no change with TrueType fonts. They are still treated as CID fonts. At any rate, there's some infrastructure available now that would make handling TrueType with multiple single byte encodings easier. Jeremias Maerki
