Tyler Minard

I've been working on it. I've had to make pretty substantial changes, though.... I'll be submitting it as a pull request and we can go from there.

This is an issue for us as well. With long running connections it appears the prepared statement cache can be poisoned and/or the cache is not accounting for plan invalidation...

We also need this to leverage deferrable constraints in postgres (indexes can never be deferrable).