io_write (io.c) bug (and its fix) under MS Windows for GUI apps (rubyw)

Hi all,
I recently encountered a problem under MS Windows where a GUI
application that runs correctly with ruby.exe refused to start with
rubuw.exe (1.8.5 version). After some digging I found that the problem
came from the function io_write in io.c : rb_sys_fail was called after
receiving -1 from io_fwrite. Actually, the ruby interpreter wanted to
print some warning to "rb_stderr" and because its a GUI application
with no console attached to it the C runtime returned -1. If I use
ruby.exe instead, the warnings are printed correctly.
To correct this behavior, it's sufficient to test if we have a console
attached to the current process and in case no one is found, we don't
call rb_sys_fail if the target stream is either rb_stderr ou
rb_stdout.
I made the following modifications to implement this :

    - Add a function named "ProcessHasConsole" in case we are
compiling for MS Windows that returns TRUE if a console is attached to
the current process and FALSE otherwise. You can find its body at the
end of this email.

    - In "io_write" body, replace the statement "if (n == -1)
rb_sys_fail(fptr->path);" with the following :
               if (n == -1)
               {
                  #ifdef _WIN32
                  if( (io != rb_stderr && io != rb_stdout) ||
ProcessHasConsole())
                  #endif
                  rb_sys_fail(fptr->path);
               }

I hope this can help Windows users of ruby that were affected by the
same problem as me.
Is there any specific protocol for submitting patches?

Cheers,

Mounir IDRASSI
IDRIX
web: http://www.idrix.fr

PS: body of the function  "ProcessHasConsole" :

#ifdef _WIN32
static BOOL ProcessHasConsole()
{
       TCHAR szTitle[128];
       if(GetConsoleTitle(szTitle,128) || GetLastError() == ERROR_SUCCESS)
                return TRUE;
       else
                return FALSE;
}
#endif