2017-10-06 21:36 GMT+02:00 Nico Williams <n...@cryptonector.com>: > On Fri, Oct 06, 2017 at 08:51:53PM +0200, Pavel Stehule wrote: > > 2017-10-06 20:39 GMT+02:00 Nico Williams <n...@cryptonector.com>: > > > On Fri, Oct 06, 2017 at 06:37:57PM +0200, Pavel Stehule wrote: > > > > When we talked about this topic, there are two issues: > > > > > > > > a) probably not too hard issue - some internal data can be in > session sys > > > > cache. > > > > > > > > b) the session sys data should be visible on SQL level too (for some > > > tools > > > > and consistency) - it is hard task. > > > > > > Can you expand on this? > > > > If global temporary tables should be effective, then you have not have > > modify system catalogue after creating. But lot of processes requires it > - > > ANALYZE, query planning. > > But the nice thing about them is that you need only create them once, so > leave them in the catalog. Stats about them should not be gathered nor > stored, since they could be different per-session. >
Unfortunately one field from pg_class are not static - reltuples should be per session. But it can be moved to different table Regards Pavel
