On Mon, Oct 21, 2024, at 11:42, Alvaro Herrera wrote: > ... actually, the "AND pg_shdepend.dbid = ( SELECT pg_database_1.oid > ...)" part of this is useless, because you already had that in the ON > condition of the original join to pg_database. So, apologies for the > noise there. TBH I don't see why you put that in the JOIN ON condition > instead of WHERE, but anyway you don't need to add a new condition for > it. I guess I'd do it like this for clarity: > > FROM pg_catalog.pg_shdepend > JOIN pg_catalog.pg_database ON pg_database.oid = > pg_shdepend.dbid > JOIN pg_catalog.pg_authid ON pg_authid.oid = > pg_shdepend.refobjid > LATERAL pg_catalog.pg_identify_object(pg_shdepend.classid, > pg_shdepend.objid, pg_shdepend.objsubid) AS identify, > LATERAL > pg_catalog.aclexplode(pg_catalog.pg_get_acl(pg_shdepend.classid, > pg_shdepend.objid, pg_shdepend.objsubid)) AS aclexplode > WHERE pg_shdepend.deptype = 'a' AND > pg_shdepend.refclassid = 'pg_catalog.pg_authid'::pg_catalog.regclass > AND > pg_database.datname = pg_catalog.current_database() > > ... but since these are inner joins, this might be a matter of style. > (I did add a couple of schema-qualifications there.)
Ahh, right, that's nicer, thanks for fixing. New patch attached. I also fixed pg_ownerships in the same way, moving the `pg_catalog.pg_database.datname = pg_catalog.current_database()` to the WHERE clause instead. /Joel
v5-0001-Add-pg_ownerships-and-pg_privileges-system-views.patch
Description: Binary data
