On 2020-Apr-20, Alvaro Herrera wrote:
> + while (HeapTupleIsValid(trigtup = systable_getnext(scan)))
> + {
> + Form_pg_trigger pg_trigger = (Form_pg_trigger) GETSTRUCT(trigtup);
> + ObjectAddress trig;
> +
> + /* Ignore triggers that weren't cloned */
> + if (!OidIsValid(pg_trigger->tgparentid) ||
> + !pg_trigger->tgisinternal ||
> + !TRIGGER_FOR_ROW(pg_trigger->tgtype))
> + continue;
Actually, shouldn't we be checking just "!OidIsValid(pg_trigger->tgparentid)"
here? Surely the other two conditions should already not matter either
way if tgparentid is set. I can't see us starting to clone
for-statement triggers, but I'm not sure I trust the internal marking to
remain one way or the other.
--
Álvaro Herrera https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services