Ildar Musin <i.musin@postgrespro.ru> writes:
> [ ri_triggers_v2.patch ]
Pushed with two minor improvements. I noticed that ri_setdefault could
just go directly to ri_restrict rather than call the two separate triggers
that would end up there anyway; that lets its argument be "TriggerData
*trigdata" for more consistency with the other cases. Also, this patch
made it very obvious that we were caching identical queries under hash
keys RI_PLAN_RESTRICT_DEL_CHECKREF and RI_PLAN_RESTRICT_UPD_CHECKREF,
so we might as well just use one hash entry for both cases, saving a few
lines of code as well as a lot of cycles. Likewise in the other two
functions.
regards, tom lane