I think I've made some progress on this.
I investigated the possibility of using Fontforge to script the copying
of glyphs from one font to another, and it turns out that this is quite
easy.
I have written two scripts:
- A Fontforge script which copies a single glyph from one font to
another.
- A shell script which copies Serbian glyphs for each of the broken
fonts.
The list of broken font files is based upon the information in the
original bug report about which fonts have problems. Since I can't read
Cyrillic, the only glyphs that are copied are
- small Tshe (aka U+0452, afii10099), because it was mentioned in the
original report, and
- large Tshe (aka U+0402, afii10051), because I'm guessing that that's
wrong too, due to the fact that it's different in Valek's version.
It should be trivial to make the shell script copy other glyphs across,
or to copy between additional fonts. If somebody can provide me with a
comprehensive list of the glyphs that are broken, this would be very
useful.
The script also uses the afmutil.py script you mentioned to generate a
diff between the AFM file before the copy and the same file afterwards,
ignoring changes in comments and such.
I ran the scripts on the fonts in the gsfonts package to copy glyphs
from the fonts in
ftp://ftp.gnome.ru/fonts/urw/release/urw-fonts-1.0.7pre40.tar.bz2
I saw some side-effects to the metrics of the scircumflex and
jcircumflex characters, but only very small ones (i.e. two numbers being
changed by 1). I'm guessing these might be due to a rounding error in
Fontforge or something like that. Other than that, the only changes are
to comments and to the metrics of the glyphs in question.
I don't know if hinting has been affected.
I have uploaded the two scripts I wrote, afmutil.py (for convenience), a
a HTML page for testing the fonts, a screenshot of that page on my
system, and a Debian package with the modified fonts, to the following
location:
http://muse.19inch.net/~daf/dump/271427/
In the screenshot, Schoolbook and Palladio look fine. Some fonts don't
appear to have bold variants. Nimbus Roman seems to have a buggy italic
variant, and Bookman has buggy glyphs for all variants, though small
Tsche seems fine for the Roman variant. The possibility that some of the
old versions of the fonts might still be hanging around in memory or
something like that, might account for some of these problems, though.
--
Dafydd
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
