CLONE 5427 - QTextCodec::codecForLocale()->fromUnicode() produces Latin-1 on UTF-8 locales on Linux

QTextCodec::codecForLocale()->fromUnicode() produces Latin-1 on UTF-8 locales on Linux

Tested two versions of gcc:
gcc (Gentoo 4.3.2-r3 p1.6, pie-10.1.5) 4.3.2
gcc (Ubuntu 4.3.3-5ubuntu4) 4.3.3

And OpenSUSE 11.4 + qt 4.7.1. As a slight variation, the bug was found out with toLocal8Bit(), but the cause is obviously the same.

Steps to reproduce / test case
QTextCodec::codecForLocale()->fromUnicode() seems to be producing Latin-1 on UTF-8 based locales (tested on en_US.UTF-8 and fr_FR.UTF-8) on Linux. The following program demonstrates the problem:

#include <QFile> 
#include <QTextCodec> 

int main(int, char **) { 
QByteArray array = QTextCodec::codecForLocale()->fromUnicode(QString::fromUtf8("\xc3\xa9")); 
QFile out(QString::fromUtf8("out")); 
Q_ASSERT(out.open(QIODevice::WriteOnly | QIODevice::Truncate)); 
Q_ASSERT(out.write(array) == array.size()); 
return 0; 
} 

Obtained output:

The file "out" contains a single byte: e9.

Expected output:

The file "out" should contain the two-byte sequence c3 a9.

Remarks:

• e9 and c3 a9 are the encodings of "é" in Latin-1 and UTF-8, respectively.
• Replacing codecForLocale() with codecForName("UTF-8") in the program yields the expected output.
  More information
  Tested on
  Qt 4.5.0-0ubuntu4.2 on Ubuntu
  Qt 4.5.1 with iconv support on Gentoo