Re: can't write wide character/string to outputfile

[Axel Freyn <axel-freyn at gmx dot de>]

Hi,
>
>
> I have slightly modified the program
>
> #include <iostream>
> #include <fstream>
> #include <string>
> using namespace std;
> int main() {
>  wstring ws2 = L"Euro:\x20ac";
>  wofstream out("unicode.txt");
> //  out.imbue(locale("el_GR.UTF-8"));
>  out << ws2<< endl;
>  wcout << ws2;
> }
>
> and this prints the word Euro on the terminal while the
> output file is still empty.
>

Well, I'm not an expert in localization, however I would propose you try
a code like the following:

#include <iostream>
#include <fstream>
#include <string>
using namespace std;
int main(int argc, char *argv[]) {
 wstring ws2 = L"Euro: \x20ac";
 wofstream out("unicode.txt");
 if(not out.good())
   cerr << "Error opening output file" << endl;
 const char *name = "el_GR.UTF-8";
 if(argc == 2)
   name = argv[1];
 cout << "trying to access locale " << name << endl;
 locale loc;
 try{
   loc = locale(name);
   cout << "Generated locale " << loc.name() << endl;
 }catch( exception &e){
   cerr << "Couldn't generate locale " << name << ": " << e.what() << endl;
 }
 out.imbue(loc);
 if(not out.good())
   cerr << "Error when setting the locale" << endl;
 out << ws2<< endl;
 if(not out.good())
   cerr << "Error when writing to file" << endl;
}

It has much more error-checking integrated, such that I would hope we
get a clear answer where the error happens. My code takes 1 command-line
parameter, which it interpretes as the desired locale.

I expect the following behaviour (let's call the compiled program "test":

 - if you start it as "./test C" or "./test POSIX", it should succeed to
   generate those locales (they are defined by the C++-Standard), but
   fail in the last line when trying to write the Euro-symbol into the
   file.

 - if you start it was an utf-8--encoding, (like el_GR.UTF-8),
   everything should run fine IF this encoding exists on your machine.
   However, the C++-standard says that the encodings are implementation
   defined, and the gcc-documentation says it's up to glibc:
   http://gcc.gnu.org/onlinedocs/gcc/Locale_002dspecific-behavior-implementation.html
   On my machine, the names are case-sensitive -- so maybe you should
   try some other spellings like el_GR.utf-8 or el_GR.utf8 or ...?

Axel