The use of PyW32_BEGIN_ALLOW_THREADS and PyW32_END_ALLOW_THREADS

[sam]

The follllowing code lifted from Mark Hammond Pywin32 code shows and
example of calling the Windows Kernel32 GetTickCount(),using
PyW32_BEGIN_ALLOW_THREADS and PYW32_END_ALLOW_THREADS. My Code does not
use this,but uses  SetThreadAffinityMask(GetCurrentThread(),1). My
questions are:

1) What is the difference?
2) What happens after I leave my function, to the previous threads that
I may have had running?
3) Should I be using the BEGIN/END ALLOW_THREADS?


/ @pymethod string|win32api|GetTickCount|Returns the number of
milliseconds since windows started.
static PyObject *
PyGetTickCount(PyObject * self, PyObject * args)
{
	if (!PyArg_ParseTuple (args, ":PyGetTickCount"))
		return NULL;
	PyW32_BEGIN_ALLOW_THREADS
	DWORD count = GetTickCount();
	PyW32_END_ALLOW_THREADS

	return Py_BuildValue("l",(long)count);
}

Now my code uses the Microsoft
"SetThreadAffinityMask(GetCurrentThread(),1)"
//
// PROGRAMMER:  Samuel W. Schulenburg
// DATE:        May 18,2006
//
// FUNCTION:    double dMyClock()
//
// DESCRIPTION: This function will return the seconds since the system
was started
//              as a double. If a call to QueryPerformanceFrequency()
returns a true
//              then QueryPerformanceCounter() will be used else the
standard 'C'
//              dMyClock() function will be used.
//
// PARAMETERS:  None
//
// RETURN:      double  The seconds passed sense the system was started
//
double dMyClock()
{
   double dLow;
   double dHigh;

   if(iPerformanceClocks == 1) // if the hardware supports the high
performance clock
     {
        SetThreadAffinityMask(GetCurrentThread(),1);
        QueryPerformanceCounter(&liPerformanceCount);
        dLow  = (double)liPerformanceCount.LowPart;
        dHigh = (double)(liPerformanceCount.HighPart)*(4294967296.0);
	return((dLow+dHigh)/dTicksPerSecond);
     }
   else // use the standard C clock function
        return((double) (clock()/dTicksPerSecond));
}