On Mon, Aug 15, 2011, Stewart, Robert wrote:
You're proposing a `text` class that is meant to somehow overcome this problem. So you change the boost interface to accept `text` but user code is left unchanged...:
namespace boost { void func(const text& a); } boost::func(non_utf_string); //oops, the std::string default constructor is called.
Yes, you can make this constructor explicit, so the above code stops compiling and the user must write explicitly: boost::func(text(non_utf_string));
But then there is nothing in your proposal that makes std::string utf-8 encoded by 'default'. Default == implicit.
As soon as the client did a cast, the client made the claim that non_utf_string met the requirements of the text class' constructor. The problem is that of the client misusing the class by an ill-advised cast. What's more, I think Soares indicated a debug-build validation that the argument indeed was UTF-8.
For my design the programmer must explicitly choose an encoding in the template instance and also explicitly call the constructor to construct the Unicode string. As long as we make it hard enough that programmer has to consciously call the constructor then the rest of the responsibility will fall on the programmer. Remember, only prevent accidental not intentional misusage.