On Tue, Jul 19, 2011 at 9:24 PM, Heikki Linnakangas
<heikki.linnakangas@enterprisedb.com> wrote:
>> You might persuade me to do it another way, but I can't see how to
>> make that way work. Your case seems a stretch.
>
> You get coincidences with memory allocations surprisingly often, because
> things tend to get allocated and free'd in chunks of certain sizes. It's
> also pretty fragile in the face of future development. It's not hard to
> imagine someone adding code in lock.c to dereference the pointer.
Then I think we need a 4th phase (would actually happen first).
I will revoke and rework.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Training & Services