Re: POD function static initialization from function return, thread

> The guarantee for local statics is not quite as strict.
>
> C++ Standard 6.7p4: "A local object of POD type (3.9) with static
> storage duration initialized with constant-expressions is initialized
> before its block is first entered. "
>
> Every compiler I've seen does this (for PODs only) by initializing
> before the program begins, with constants in the data
> segment. Non-PODs are initialized "the first time control p*****
> through its declaration", which is usually unsynchronized, and
> is therefore dangerous with multiple threads.

So that begs the question of whether static values in initialization
lists are done before first entry into the function. I prefer the
former idiom against the latter, but the latter appears to be "more
guaranteed" to be thread-safe:
class Mutex
{
  pthread_mutex_t _mutex;
  Mutex() : _mutex(PTHREAD_MUTEX_INIT) {}
};
void Foo()
{
  static Mutex mutex;
}

The following appears to always be safe.
class Mutex
{
  pthread_mutex_t _mutex;
};
#define MUTEX_INIT PTHREAD_MUTEX_INIT

void Foo()
{
  static Mutex mutex = MUTEX_INIT;
};

Am I expecting too much out of my compilers?

-Brian