Glib::ustring and Japanese characters

Question

Glib::ustring is supposed to work well with UTF8 but I have a problem when working with Japanese strings.

If you compare those two strings, "わたし" and "ワタシ", using == operator or compare method, it will answer that those two strings are equals.

I don't understand why. How Glib::ustring works ?

The only way I found to get false to the comparison is to compare strings of different sizes. For example "海外わたわ" and "海外わた".

Very strange...

Answer 1

Glib::ustring::compare uses g_utf8_collate() internally, which compares strings according to the rules of the current locale. Is your locale set to something other than Japanese?

Answer 2

#include <iostream>
#include <glibmm/ustring.h>
int main() {
  Glib::ustring s1 = "わたし";
  Glib::ustring s2 = "ワタシ";
  std::cerr << (s1 == s2) << std::endl;
  return 0;
}

Output: 0

EDIT: But I digged a little deeper:

#include <iostream>
#include <glibmm.h>
int main() {
  Glib::ustring s1 = "わたし";
  Glib::ustring s2 = "ワタシ";
  std::cout << (s1 == s1) << std::endl;
  std::cout << (s1 == s2) << std::endl;
  std::locale::global(std::locale(""));
  std::cout << (s1 == s1) << std::endl;
  std::cout << (s1 == s2) << std::endl;
  std::cout << s1 << std::endl;
  std::cout << s2 << std::endl;
  return 0;
}

Output:

1
0
1
1
わたし
ワタシ

And this sounds strange.