mistercat saidcode fragment in the article about string reversing is just wrong. there is 1) an overflow error, 2) complete misuses of c++ programming language for the particular problem.
Thanks for your comment mistercat.
For the overflow, do you mean the int should be a std::string::size_type? That won't necessarily cause an overflow but I think we should certainly class it as a bug nonetheless.
I agree that the solution is clever, but misguided. The problem is that std::basic_string is only suitable for use with writing systems where each char or wchar_t is a complete character. For Unicode this means that we have to make sure that the character type we use is 32 bits wide. Normally a massive waste of computer memory. Other encoding systems can't be reliably manipulated at all. Another reader sent me a link to a post by Nathan Myers describing the limitation on the C++ string types.
In FOST.3™ we've ducked around it by implementing a FSLib::wstring that knows it is a UTF-16 sequence and behaves accordingly. The interface is (almost) identical to that for std::wstring.
