[cjk] filed two more issues, and have a workaround (Re: AW: Re: Argh I see, dvips seems to be producing rubbish!)

[Hin-Tak Leung]

I filed two more issues at the fontforge issue tracker:

https://github.com/fontforge/fontforge/issues/3084
https://github.com/fontforge/fontforge/issues/3085

Anyway, I have a workaround. Here is a summary of the two issues:

- the main reason, and a rather interesting one, why Thorsten suffers less from 
mis-encoding, is because his fontforge installation is incomplete :-). Ubuntu 
ships the cidmap part of fontforge as a separate extra package 
"fontforge-extras", whereas on fedora, it is just one for the whole thing. Once 
I move/hide the Adobe-Identity cmap, I got rid of the mis-encoding too.

- I can work around the missing glyph issues by manually duplicating the 
glyphs. I have a proof-of-concept subfont.pe script which does it for a few 
missing glyphs. However, fontforge does not give scripting access to the 
forward/backend encoding tables, so basically the process involves reading and 
keeping the encoding/decoding table by another means, do the re-encode, fix up 
the missing glyphs, before writing the subfonts.

With 400 glyhs to fix up, I am going to need to write a script to do that, 
reproducibly... it is probably going to be python-based, using freetype-py ; 
the reason is that fontforge already have a python-based althernative scripting 
interface as the *.pe native script style, so it would be a good thing 
eventually to migrate the subfont.pe native script to a python-based one, once 
fontforge's scripting interface provides access to the encoding/decoding tables.

So Thorsten, you'll need to install freetype-py (or python2/3-freetype; don't 
know what it is called under Ubuntu) in anticipation of that.