Ok, I'll give wxWidgets some more thought and investigation.
About Python, I wasn't thinking to implement Qi in Python, but rather
to initially embed standard Python in an experimental Qi system, just
as it is embedded in say Blender as a scripting language. Of course Qi
doesn't need a scripting language, but it would allow using any Python
library from Qi. Pure python would be slow, but with a bit more work C
module extensions of python could be called more directly, bypassing
the Python interpreter, so they would be faster. And of course Qi
would still be as fast as it already is.
If that succeeded and was useful, Python could be implemented in Qi,
not embedded, and with no interpreter, but rather through a translator
written in Qi which would make use of the underlying CL compiler as
the last step. Then all Python could run pretty fast, as long as the
translator was written to be fast for when Python does an eval, but
from what I've read, Qi can easily handle it. I bet a lot of Python
users would love it just to run Python fast. It could get a lot of
publicity for Qi.
The question though is: would that necessarily ruin the Qi manifesto
just to get libraries?