Re: Strings and bindings

[Darin Adler <darin eazel com>]

> [Owen discusses UTF-8, non-UTF-8, std::wstring, and std::string for GTK+]

If GTK+ is using UTF-8 for strings, there's no reason to mess with the
std::wstring type; a char * is naturally represented with a std::string,
whatever the encoding. There's no reason to tie the conversion from UTF-8 to
a fixed-width encoding to the translation across the C to C++ boundary.

> So, the question is, do we need two types in the type system for
> user-visible and non-user-visible strings or just one? My default
> answer is that we should keep it simple, and just have one, but I'm
> very willing to accept input on this issue.

Besides the issue for the Python binding, the only issue is keeping straight
what the encoding is for each string. If non-user-visible strings are in
ASCII, that's no problem since it's a proper subset of UTF-8. So, I think
that an "all UTF-8" strategy is a great idea.

    -- Darin