Hey guess what, we need to deal with Unicode strings in Python too! Unicode 
strings are a different type than normal strings.  Swig easily maps 'some 
to char* or std::string, but doesn't seem to want to map u'some string'.  You 
have to convert it with string.encode("ascii") or something like that.
Might Swig map u'some string' to wchar_t* or wstring? I'll have to try it and
find out.


Been thinking a bit more, I guess we could say that in general, textual Property
data should be only utf-8 and no other encoding.  Certain textual property 
datatypes restrict the characters allowed anyway, for instance ints and floats,
and the characters those are restricted to would be plain ascii compatible
anyway.  Then property just needs wstring methods I guess.

I'm still not sure about object names and URLs.  I have a feeling that we 
hold off on that. That could get pretty complicated.

Is there any reason to support other charsets like the windows encodings or
the "koi" encodings or whatever-- do they offer anytheng that unicode doesn't 
(other than existing documents in those charsets)?


vos-d mailing list

Reply via email to