On Sun, Jul 31, 2005 at 11:47:55AM +0200, Martin Husemann wrote:
> Out of curiosity, care to give any pointers/details/implementation details?

Consider a system that uses the x86 segmented model and uses separate
descriptors for each malloc'ed area. Loading a segment register with
a stale pointer will cause a protection error.

> Withoug looking at the standard, my gut feeling is that those implementaitons
> would be non conforming, but it would be very entertaining to be proven
> wrong here ;-)

I read recently that a conforming implementation could give you the
next index in a lookup table for each new pointer you create......

	David

-- 
David Laight: david@l8s.co.uk