Comment 11 for bug 1649066

Non-Windows versions of Text::utf8ToWide and Text::wideToUtf8 won't obviously handle UTF-16 surrogate pairs at all, thus producing incorrect results. See the following test case for 🌍:

string toUtf8(const wstring& str) {
 string tgt;
 string::size_type n = str.length();
 for (string::size_type i = 0; i < n; ++i) {
  Text::wcToUtf8(str[i], tgt);
 }
 return tgt;
}

wstring emoji = L"\U0001F30D";
ASSERT_EQ(Text::wideToUtf8(emoji), toUtf8(emoji)); // error