Simon Riggs <[EMAIL PROTECTED]> writes: > [ ... a large amount of analysis based on exactly one test case ... ]
I think you are putting too much emphasis on fixing one case and not enough on considering what may happen in other cases ... In general, estimating n-distinct from a sample is just plain a hard problem, and it's probably foolish to suppose we'll ever be able to do it robustly. What we need is to minimize the impact when we get it wrong. So I agree with the comment that we need to finish the unfinished project of making HashAggregate tables expansible, but I'm dubious about the rest of this. regards, tom lane ---------------------------(end of broadcast)--------------------------- TIP 9: In versions below 8.0, the planner will ignore your desire to choose an index scan if your joining column's datatypes do not match