Vitalii wrote
>
> Since cte is already an optimization fence, you can go further and make
> it temporary table.
> Create table;analyze;select should make optimizer's work much easier.
>
Thanks Vitalii - yes, you are right, and I have used that technique on
other cases like this.
How
On 2012-10-09 23:09:21
Tom Lane wrote:
>
> re subject Why am I getting great/terrible estimates with these CTE queries?
> You're assuming the case where the estimate is better is better for a
> reason ... but it's only better as a result of blind dumb luck. The
> outer-level query planner do
