Re: [cjk] Argh I see, dvips seems to be producing rubbish!

[Hin-Tak Leung]

--------------------------------------------
On Thu, 8/6/17, Werner LEMBERG <address@hidden> wrote:
 
> The `missing glyphs' are glyph variants. 
 Usually, glyph X is, say,
 the Japanese
 variant, while X+1 is the Taiwanese one, etc., etc.


Argh, I see.  cmap (3,10) have about 400 pairs of unicode points maps to the 
same glyph. The re-encoding process is likely dropping one of them. It would 
make sense to duplicate or share a link...

> Maybe the bug can be circumvented by adjusting
 the .pe script so that
 it temporarily
 removes all glyphs that don't have a cmap mapping? 
 I
 haven't the fontforge docs whether
 this is possible...
 

It probably is. I see that after reencoding, I have about 49000 characters (up 
from 43000), so about 6000 unicode points are being trashed.

I am in the middle of changing stuff in both fontforge and the subfont script 
at the moment. ttf2tfm has a easier task: it just needs to point to the right 
glyph, so sharing a glyph between two code points is less an issue, compared to 
the case with a font editor.