On Mon, Mar 18, 2024 at 6:01 PM jian he <jian.universal...@gmail.com> wrote: > > On Mon, Mar 18, 2024 at 3:09 PM Peter Eisentraut <pe...@eisentraut.org> wrote: > > > > On 14.03.24 01:26, Michael Paquier wrote: > > > -EventTriggerSupportsObjectClass(ObjectClass objclass) > > > +EventTriggerSupportsObject(const ObjectAddress *object) > > > > > > The shortcut introduced here is interesting, but it is inconsistent. > > > HEAD treats OCLASS_SUBSCRIPTION as something supported by event > > > triggers, but as pg_subscription is a shared catalog it would be > > > discarded with your change. Subscriptions are marked as supported in > > > the event trigger table: > > > https://www.postgresql.org/docs/devel/event-trigger-matrix.html > > > > Ah, good catch. Subscriptions are a little special there. Here is a > > new patch that keeps the switch/case arrangement in that function. That > > also makes it easier to keep the two EventTriggerSupports... functions > > aligned. Also added a note about subscriptions and a reference to the > > documentation. > > select relname from pg_class where relisshared and relkind = 'r'; > relname > ----------------------- > pg_authid > pg_subscription > pg_database > pg_db_role_setting > pg_tablespace > pg_auth_members > pg_shdepend > pg_shdescription > pg_replication_origin > pg_shseclabel > pg_parameter_acl > (11 rows) >
also in function doDeletion we have: /* * These global object types are not supported here. */ case AuthIdRelationId: case DatabaseRelationId: case TableSpaceRelationId: case SubscriptionRelationId: case ParameterAclRelationId: elog(ERROR, "global objects cannot be deleted by doDeletion"); break; do we need to add other global objects? in the end, it does not matter since we have: default: elog(ERROR, "unsupported object class: %u", object->classId);
