The Unicode Consortium Discussion Forum

I was recently tasked to add unicode support for a C++ toolkit, a C++ library that is able to run on different platforms
1. In the existing C++ library, the string class can handle both char and wchar_t, string and wstring ( The implementation using Template )
2. As I mentioned before, the C++ library is capable of of running on a variety of different platforms such as Windows. Linux, Unix ( AIX, Solaris, HP etc)
3. after some google search. Right now I am leaning towards using uft-8 encoding for the C++ library.

My question here: Any suggestions on what approach I shall take?

Technically, the wchar_t should be set to the widest code unit, so char as byte and string as utf-8 would work, and then wchar_t would be UTF32.

However, many implementation find it practical to use UTF16 in wchar_t with the downside that surrogate pairs aren't automatically handled, but surfaced to the user of the library. The upside is that on many plaforms, UTF-16 is the native format....