Re: support for MERGE
Alvaro Herrera <alvherre@alvh.no-ip.org>
From: Álvaro Herrera <alvherre@alvh.no-ip.org>
To: Justin Pryzby <pryzby@telsasoft.com>
Cc: Peter Eisentraut <peter.eisentraut@enterprisedb.com>, Amit Langote <amitlangote09@gmail.com>, Japin Li <japinli@hotmail.com>, Zhihong Yu <zyu@yugabyte.com>, Simon Riggs <simon.riggs@enterprisedb.com>, pgsql-hackers@lists.postgresql.org, Tomas Vondra <tomas.vondra@enterprisedb.com>, Daniel Westermann <dwe@dbi-services.com>, Erik Rijkers <er@xs4all.nl>, Jaime Casanova <jcasanov@systemguards.com.ec>, Andres Freund <andres@anarazel.de>
Date: 2022-08-01T15:30:30Z
Lists: pgsql-hackers
On 2022-Aug-01, Justin Pryzby wrote:
> On Tue, Jun 14, 2022 at 11:27:06AM +0200, Álvaro Herrera wrote:
> > @@ -448,9 +448,9 @@ COMMIT;
> > and execute the first one that succeeds.
> > If <command>MERGE</command> attempts an <command>INSERT</command>
> > and a unique index is present and a duplicate row is concurrently
> > - inserted, then a uniqueness violation is raised.
> > - <command>MERGE</command> does not attempt to avoid the
> > - error by executing an <command>UPDATE</command>.
> > + inserted, then a uniqueness violation error is raised;
> > + <command>MERGE</command> does not attempt to avoid such
> > + errors by evaluating <literal>MATCHED</literal> conditions.
>
> This was a portion of a chang that was committed as ffffeebf2.
>
> But I don't understand why this changed from "does not attempt to avoid the
> error by executing an <command>UPDATE</command>." to "...by evaluating
> <literal>MATCHED</literal> conditions."
>
> Maybe it means to say "..by re-starting evaluation of match conditions".
Yeah, my thought there is that it may also be possible that the action
that would run if the conditions are re-run is a DELETE or a WHEN
MATCHED THEN DO NOTHING; so saying "by executing an UPDATE" it leaves
out those possibilities. IOW if we're evaluating NOT MATCHED INSERT and
we find a duplicate, we do not go back to MATCHED. We have this comment
in ExecMerge:
* ExecMergeMatched takes care of following the update chain and
* re-finding the qualifying WHEN MATCHED action, as long as the updated
* target tuple still satisfies the join quals, i.e., it remains a WHEN
* MATCHED case. If the tuple gets deleted or the join quals fail, it
* returns and we try ExecMergeNotMatched. Given that ExecMergeMatched
* always make progress by following the update chain and we never switch
* from ExecMergeNotMatched to ExecMergeMatched, there is no risk of a
* livelock.
(Another change there is the period to semicolon. I did that to make it
clear that the last phrase applies to only that part and not phrases
earlier in the same paragraph.)
Your proposed rewording might be a clearer way to express the same idea.
Any other opinions?
> Sorry to re-raise this 6 weeks later..
No worries. As the Zen of Python says, now is better than never.
--
Álvaro Herrera 48°01'N 7°57'E — https://www.EnterpriseDB.com/
"All rings of power are equal,
But some rings of power are more equal than others."
(George Orwell's The Lord of the Rings)
Commits
-
Doc fixes for MERGE statement
- 5dcebd10baa8 16.0 landed
- 5bb88967ed5c 15.0 landed
-
MERGE docs adjustments
- d6d1fbf353d1 15.0 landed
- 3d895bc846f2 16.0 landed
-
Link to MVCC docs in MERGE docs
- ffffeebf2447 15.0 landed
-
Fix EXPLAIN MERGE output when no tuples are processed
- 12e423e21d8e 15.0 landed
-
Make EXPLAIN MERGE output format more compact
- 598ac10be1c2 15.0 landed
-
Add support for MERGE SQL command
- 7103ebb7aae8 15.0 landed
-
Split ExecUpdate and ExecDelete into reusable pieces
- 25e777cf8e54 15.0 landed
-
Add API of sorts for transition table handling in trigger.c
- 3a46a45f6f00 15.0 landed
-
Represent Lists as expansible arrays, not chains of cons-cells.
- 1cff1b95ab6d 13.0 cited