Tom Lane wrote:
> The postmaster's dependence on the contents of shared memory should
> ideally be zero (and it is zero, or nearly so, at the moment).
> Otherwise a backend crash that clobbers shared memory poses the risk
of
> taking down the postmaster as well. We can't go in that direction.
Just a thought: win32 memory mapped files can be designated read only at
the time of the 'mapping'. This would guarantee the memory in question
would not be clobbered by anybody but the postmaster. Also, the MMF for
the data in question could be kept separate from the general pool by
giving it a separate filename. This is a fairly elegant approach from
win32 point of view. Not sure if you are trying to carry this to the
unix side, which of course would make this moot.
Regards,
Merlin