Question

I wrote the following program using VS2008:

#include <fstream>
int main()
{
    std::wofstream fout("myfile");
    fout << L"Հայաստան Россия Österreich Ελλάδα भारत" << std::endl;
}

When I tried to compile it the IDE asked me whether I wanted to save my source file in unicode, I said “yes, please”.
Then I run the program, and myfile appeared in my project’s folder. I opened it with notepad, the file was empty. I recalled that notepad supported only ASCII data. I opened it with WordPad, it was still empty. Finally the little genius inside me urged me to look at the file size and not surprisingly it was 0 bytes. So I rebuilt and reran the program, to no effect. Finally I decided to ask very intelligent people on StackOverflow as to what I am missing and here I am 🙂

Edited:

After the abovementioned intelligent people left some comments, I decided to follow their advice and rewrote the program like this:

#include <fstream>
#include <iostream>
int main()
{
    std::wofstream fout("myfile");
    if(!fout.is_open())
    {
        std::cout << "Before: Not open...\n";
    }
    fout << L"Հայաստան Россия Österreich Ελλάδα भारत" << std::endl;
    if(!fout.good())
    {
        std::cout << "After: Not good...\n";
    }
}

Built it. Ran it. And… the console clearly read, to my surprise: “After: Not good…”.
So I edited my post to provide the new information and started waiting for answers which would explain why this is and what I could do. 🙂

Answer 1

In Visual studio the output stream is always written in ANSI encoding, and it does not support UTF-8 output.

What is basically need to do is to create a locale class, install into it UTF-8 facet and then imbue it to the fstream.

What happens that code points are not being converted to UTF encoding. So basically this would not work under MSVC as it does not support UTF-8.

This would work under Linux with UTF-8 locale

#include <fstream>
int main()
{
    std::locale::global(std::locale(""));
    std::wofstream fout("myfile");
    fout << L"Հայաստան Россия Österreich Ελλάδα भारत" << std::endl;
}

~
And under windows this would work:

#include <fstream>
int main()
{
    std::locale::global(std::locale("Russian_Russia"));
    std::wofstream fout("myfile");
    fout << L"Россия" << std::endl;
}

As only ANSI encodings are supported by MSVC.

Codecvt facet can be found in some Boost libraries. For example: http://www.boost.org/doc/libs/1_38_0/libs/serialization/doc/codecvt.html