Thread: Template Instantiation

Code:

	basic_string<T> lol, rofl;

	if(typeid(T) == typeid(wchar_t))
	{
		lol = L"Time", rofl = L"Name";
	}

	if(typeid(T) == typeid(char))
	{
		lol = "Time", rofl = "Name";
	}

This gives compile error

error C2679: binary '=' : no operator found which takes a right-hand operand of type 'const char [5]' (or there is no acceptable conversion)

You know what I'm trying to do, wide string if the instatiation is unicode, narrowed if not. How do I express myself?

Unfortunately the _T or _TEXT macros won't work if the type T is not dependent on whether UNICODE is defined.

One possible solution is to have a templated function that has a specialization for one or more types T that you wish to make a special case for.

And _T, etc are non-portable.

The reason for the compiler error is that lol and rofl are of type basic_string<char> (essentially a raw std::string), which supports an assignment operator that accepts a char * but not an assignment operator that accects a wchar_t * argument. For that case, those variables need to be of type basic_string<wchar_t>.

You need to make up your mind which type you want your variables to be.

I wish folks would just stop using TCHAR's.....

Are you actually writting code that needs to run on Windows 95? If not, just use wchar_t and be done with it.

gg

I am writing a program to convert make m3u/pls out of the text files itunes exports. It doesn't export regular playlists.

The text file exported is in uniucode but I have some serious froblems with the wifstream so I was just converting it to ascii. I will try to articulate my problems with the wifstream, but until then I made an ugly hack.

If I were to use TCHAR's, then maybe it would generate a binary for processing ASCII files and then another for UNICODE files. Meh.

Incidentally, typeid is also a run time operator. For this sort of thing to compile, you need to use template specialisation.

Code:

template<class T> class X
{
     public:
            X() {};
            void set();

     private:
          std::basic_string<T> x;
};

template<class T> void X<T>::set() {}

template<> void X<char>::set()
{x = "Hello";}

template<> void X<wchar_t>::set()
{x = L"Hello";}

int main()
{
      X<char> c;
      c.set();
      X<wchar_t> d;
      d.set();
}

What you have to remember when streaming Unicode (at least with the MSCRT) you have to open the stream in binary mode. Which also means that "\r\n" sequences are not auto-magically converted to "\n" for you. You also have to deal with any BOM that might be at the front of the file - which would tell you if you need to deal with any endianess issues. And since wchar_t is 2 bytes, a wstream is really only good for dealing with UTF-16.

http://msdn2.microsoft.com/en-us/lib...8e(VS.80).aspx

For a list of BOM's: http://www.i18nguy.com/unicode/c-unicode.html#BOM

gg

That's pretty clever grumpy, but like, I need to have it be extensible to accommodate an arbitrary string (not just ?hello?)

Maybe a struct would be better then? You could always package together the functionality you need and then construct strings of the proper width.

It would be simpler to work with in that case perhaps, because it could always be a member to some bigger class or whatever you have planned for design.

Code:

#include <string>

template < typename charT > 
struct bar
{
    std::basic_string< charT > lol, rofl;

    bar ( const charT * time, const charT * name  ) :
        lol( time ), rofl( name )
    {
    }

};

int main ( )
{
    bar< char > a( "Name", "Time" );
    bar< wchar_t > b( L"Name", L"Time" );

    a.lol = "example";
    b.lol = L"example";
}

But this is just an example; I'm sure there's a tinker to grumpy's solution that works.