[Mingw-notify] [ mingw-Bugs-1030849 ] Characters broken for Unicode

[SourceForge.net <no...@so...>]

Bugs item #1030849, was opened at 2004-09-20 08:55
Message generated for change (Comment added) made by dannysmith
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102435&aid=1030849&group_id=2435

Category: gcc
Group: Feature requests
>Status: Closed
>Resolution: Fixed
Priority: 5
Submitted By: Gabriel Linder (mingwlinder)
Assigned to: Danny Smith (dannysmith)
Summary: Characters broken for Unicode

Initial Comment:
Windows XP SP2
gcc version 3.4.1
ld version 2.15.91 20040904
mingw-runtime 3.5
w32api 3.1

Compiling the code below return this error :
main.c:23:1: converting to execution character set:
Illegal byte sequence.

Replacing the '°' with a space ' ' works fine, but I
need some characters like this '°'... And the same code
compile fine on Visual C++ 7.1.3088 !

// begin sample code
#define UNICODE
#define _UNICODE
#define WIN32_LEAN_AND_MEAN

#include <windows.h>
#include <commctrl.h>
#include <shellapi.h>

int main(void)
{
  LPTSTR lpCmdLine = GetCommandLine();

  TCHAR pom[MAX_PATH];
  wsprintf(pom, TEXT("Arguments: [%s]"), lpCmdLine);
  MessageBox(NULL, pom, TEXT("Information"),
MB_ICONINFORMATION | MB_OK);

  INT argc;
  LPTSTR* argv = CommandLineToArgvW(lpCmdLine, &argc);

  INT i = 0;
  while (i < argc)
  {
    wsprintf(pom, TEXT("Argument n°%i: [%s]"), i, argv[i]);
    MessageBox(NULL, pom, TEXT("Information"),
MB_ICONINFORMATION | MB_OK);
    ++i;
  }

  return 0;
}
// end sample code

----------------------------------------------------------------------

>Comment By: Danny Smith (dannysmith)
Date: 2004-09-24 21:43

Message:
Logged In: YES 
user_id=11494

ReporterP confirms that bug is fixed with latest RC, so closing
Danny

----------------------------------------------------------------------

Comment By: Gabriel Linder (mingwlinder)
Date: 2004-09-23 02:22

Message:
Logged In: YES 
user_id=1121203

I have just tried the new GCC 3.4.2 version, and it works
fine with -finput-charset=latin1. Many thanks ! ;-)

----------------------------------------------------------------------

Comment By: Gabriel Linder (mingwlinder)
Date: 2004-09-21 11:07

Message:
Logged In: YES 
user_id=1121203

All right, I'll wait for 3.4.2 so. But will it be built with
libiconv by default, or shall I rebuild it ?

----------------------------------------------------------------------

Comment By: Danny Smith (dannysmith)
Date: 2004-09-20 18:26

Message:
Logged In: YES 
user_id=11494

Confirmed.  
With 3.4.1 one workaround is:



  INT i = 0;
  TCHAR degree = 0xb0;
  while (i < argc)
  {
    wsprintf(pom, TEXT("Argument n%c %i: [%s]"), degree, i, 
argv[i]);
    MessageBox(NULL, pom, TEXT("Information"), 
MB_ICONINFORMATION | MB_OK);
    ++i;
  }

In order to fix properly with current gcc sources is to add 
dependence on libiconv to allow LATIN1 narrow->wide charset 
conversion.

With 3.4.2 built with libiconv,  your testcase works with this 
command line switch  -finput-charset=LATIN1

Danny

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=102435&aid=1030849&group_id=2435