That would be:
string s = boost::string::convert(-1); // to string int i = boost::string::convert<int>(str); // from string
Yes, or string s = boost::string::convert(-1, default_str); // to string int i = boost::string::convert(str, -1); // 'int' deduced from the second parameter
And with that we're almost back at the interface of boost::lexical_cast once again.
Yes, in fact my very first request was to extend lexical_cast. However, it was collectively decided to start a separate library to accommodate formatting, locales, etc.
Also the user might want to convert between std::wstring and std::string, and in C++0x, std::u8string, std::u16string, and std::u32string.
Yes, now I am coming to realize that accommodating all that additional conversion-related functionality is better done without disturbing the existing lexical_cast. I am hoping the proposed boost::string::convert() "library" is sufficiently flexible to accommodate future growth including the conversions you mentioned. They will be implemented as specializations/overloads of that same boost::string::convert(). Although I have a suspicion that std::string<->std::wstring is not such an easy task. It is because (at present anyway) in my program I do not know if std::string is, say, UTF8 ot MBCS (which might happen to be UTF8 as on Linux or might not as on Windows, Solaris). With std::u8string, std::u16string, and std::u32string it certainly will be easy as we know what we are converting from/to. So, I suspect it'll be as easy as having std::wstring convert(std::u8string) std::u8string convert(std::wstring) Thanks, Vladimir.