On Wednesday 18 March 2009 17:28, Bill Oliver wrote:

> I *think* that this information is only useful to the optimizer if we
> implement HASH join, which we don't have now, I may be wrong on this.
>
> For now, it is fine that I capture the requirement of providing the MCV,
> MCF, and histogram information through monitoring tables when/if it becomes
> available. I looked for a ticket adding HASH JOIN's but didn't see it...
> Don't we have this somewhere?

Bill, I do not think we have such ticket. What we do plan is to add histogram
information to indices. This is related with execution plans, that can vary
depending upon parameters. But this is longterm, I'm far not sure we will
have it even in 3.0.