static constructors/destructors and atexit

[Noel Yap]

Ulrich Drepper wrote:
> But a few weeks ago I described how I can imagine a better mechanism
> to be implemented (better because it's faster and has no such limits)
> when you give up 100% standard compliance by not demanding that
> user-registered functions (using atexit) run between the destructors
> for the object constructed before and after the atexit call.

Yes, but I can easily see why the standard is the way it is.  If you're
not going to stick to the standard, you might as well just override the
existing atexit() function (when it's possible).

Noel