Horton

Re: Foreign keys and partitioned tables

Alvaro Herrera <alvherre@alvh.no-ip.org>

View thread View on postgr.es
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
To: Peter Eisentraut <peter.eisentraut@2ndquadrant.com>
Cc: Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2018-04-03T20:47:23Z
Lists: pgsql-hackers 
Alvaro Herrera wrote:
> While adding some more tests for the "action" part (i.e. updates and
> deletes on the referenced table) I came across a bug that was causing
> the server to crash ... but it's actually a preexisting bug in an
> assert.  The fix is in 0001.

Yeah, it's a live bug that only manifests on Assert-enabled builds.
Here's an example:

create table pk (a int, b int, c int, d int primary key);
create table fk (d int references pk);
insert into pk values (1, 2, 3, 4);
insert into fk values (4);
delete from pk;

Will fix

-- 
Álvaro Herrera                https://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services

Commits

  1. Restore erroneously removed ONLY from PK check

  2. Foreign keys on partitioned tables

  3. Don't clone internal triggers to partitions

  4. Pass correct TupDesc to ri_NullCheck() in Assert