Michael Droettboom <mdroe@...86...> writes:
All the magic happens in "convert_ttf_to_ps", which is C code, called
from backend_ps.py. I'd start by seeing if that function is even
called, and if not, why...
One possible source of platform-specific issues is
cbook.get_realpath_and_stat, which is used on the font files to obtain
hash keys. Does stat do anything sensible on Windows?
Mark: Could you try applying the following patch to your installed
version of lib/matplotlib/backends/backend_ps.py? There should be no
need to recompile anything. If you don't have a patch utility, just make
a backup of the file, open it in a text editor, find the relevant parts,
and add each line prefixed with + below (omitting the + signs,
naturally). Then run your script and post the output. That should tell
us if the chi character is not even registered as used, or if the bug is
in the embedding code.