Horton

Re: Rename of triggers for partitioned tables

Arne Roland <a.roland@index.de>

View thread View on postgr.es
From: Arne Roland <A.Roland@index.de>
To: Alvaro Herrera <alvherre@alvh.no-ip.org>
Cc: vignesh C <vignesh21@gmail.com>, Zhihong Yu <zyu@yugabyte.com>, Pg Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2021-07-22T17:22:57Z
Lists: pgsql-hackers 
From: Alvaro Herrera <alvherre@alvh.no-ip.org>
Sent: Thursday, July 22, 2021 18:20
To: Arne Roland
Subject: Re: Rename of triggers for partitioned tables

If we have good use for such an index, I don't see why we can't add it.
But I'm not sure that it is justified -- certainly if the only benefit
is to make ALTER TRIGGER RENAME recurse faster on partitioned tables, it
is not justified.

Speed is not really the issue here, most people will have under 20 triggers per table anyways. I think even the simpler code would be worth more than the speed gain. The main value of such an index would be the enforced uniqueness.

I just noticed that apparently the
ALTER TABLE [ IF EXISTS ] [ ONLY ] name [ ENABLE | DISABLE ] TRIGGER ...
syntax albeit looking totally different and it already recurses, it has precisely the same issue with pg_dump. In that case the ONLY syntax isn't optional and the 2. from your above post does inevitably apply.

Since it is sort of the same problem, I think it might be worthwhile to address it as well within this patch. Adding two to four ereports doesn't sound like scope creeping to me, even though it touches completely different code. I'll look into that as well.

Regards
Arne

Commits

  1. Remove newly added useless assertion check

  2. Make ALTER TRIGGER RENAME consistent for partitioned tables

  3. Fix pg_dump for disabled triggers on partitioned tables

  4. Preserve firing-on state when cloning row triggers to partitions

  5. Fix ALTER TABLE .. ENABLE/DISABLE TRIGGER recursion

  6. Record parents of triggers

  7. Allow FOR EACH ROW triggers on partitioned tables