4 ms·
On Postgres 12, CTE's are not optimization fences anymore. [0] [0] https://www.depesz.com/2019/02/19/waiting-for-postgresql-12-allow-user-control-of-cte-materi
by emilsedgh 7y ago
On Postgres 12, CTE's are not optimization fences anymore. [0]
[0] https://www.depesz.com/2019/02/19/waiting-for-postgresql-12-allow-user-control-of-cte-materialization-and-change-the-default-behavior/ https://www.depesz.com/2019/02/19/waiting-for-postgresql-12-...
- rosser 7y agoNot quite true. The default behavior changes, and new syntax is added allowing users to specify their desired behavior, but if the sub-query in the CTE does certain things, it will still be materialized instead, regardless. That is: they mostly aren't a fence any more, but can be when you want them to, and sometimes still must be.