I said:
> I did find that the PROC structures are not counted in the initial
> sizing of the shared memory block.
Er ... yes they are ... never mind ...
At this point I am not going to make any further changes to the
shared-mem code unless Hiroshi and Tatsuo report that there are still
problems.
I would still like to alter the calling conventions for hash_search
and hash_seq, which are ugly and also dependent on static state
variables. But as far as I can tell right now, those are just code
beautification items rather than repairs for currently-existing bugs.
So it seems best not to risk breaking anything with so little testing
time left in the 6.5 cycle. I will put them on my to-do-for-6.6 list
instead.
regards, tom lane