From: Nico Williams <nico@cryptonector.com>
To: Peter Geoghegan <pg@bowt.ie>
Cc: Simon Riggs <simon@2ndquadrant.com>
Cc: Robert Haas <robertmhaas@gmail.com>
Cc: Michael Paquier <michael.paquier@gmail.com>
Cc: PostgreSQL-development <pgsql-hackers@postgresql.org>
Subject: Re: MERGE SQL Statement for PG11
Date: Mon, 30 Oct 2017 13:36:50 -0500
Message-ID: <20171030183649.GM4496@localhost> (raw)
In-Reply-To: <CAH2-Wzm0A0PUiAeMsmj6P+P4jaP+yz3dXH5f=xPWnuyX5m8ebQ@mail.gmail.com>
References: <CA+TgmoZuce2iHSBVpkX5qud-UbxhtY85opA47LTP=QW16J47wg@mail.gmail.com>
<CANP8+j+jG76Okkibk7Xu=T6C5-gSpXX9sudANS9nGyR9yWEgRA@mail.gmail.com>
<CA+TgmobMx66NWCoje=PEGnbMQQdZojJi7v9RqsuY23vkNsTf1Q@mail.gmail.com>
<CAB7nPqRzN4bE7EfVFgt64LA8FLMCV9=VGr8=FeogVEOiUbj4Ww@mail.gmail.com>
<CANP8+j+FPEqpd8O3scwOKUgRvtWB5r5RhcCuJ8nn6VU_wc2Nng@mail.gmail.com>
<CAH2-Wz=AxEKE7sJEukPN3Vgu6sU5QwuLCBHuF-oJbkOLWsxTQw@mail.gmail.com>
<CANP8+j+FM8ep1Cprbjaeknzq1pRAaxTMvX=hYsc4eGcj13fz7w@mail.gmail.com>
<CA+TgmoYbytQT1pZDOK6t-XD=jWffDXaHoubsU-_mqJU6W0oK0w@mail.gmail.com>
<CANP8+jJLvkBgaet6O3j-yGapu9kUSv69PTMQxGwYuq2Eb-t2dw@mail.gmail.com>
<CAH2-Wzm0A0PUiAeMsmj6P+P4jaP+yz3dXH5f=xPWnuyX5m8ebQ@mail.gmail.com>
List-Unsubscribe: <mailto:majordomo@postgresql.org?body=unsub%20pgsql-hackers>
On Mon, Oct 30, 2017 at 10:59:43AM -0700, Peter Geoghegan wrote:
> On Mon, Oct 30, 2017 at 6:21 AM, Simon Riggs <simon@2ndquadrant.com> wrote:
> > If a general purpose solution exists, please explain what it is.
>
> For the umpteenth time, a general purpose solution is one that more or
> less works like an UPDATE FROM, with an outer join, whose ModifyTable
> node is capable of insert, update, or delete (and accepts quals for
> MATCHED and NOT matched cases, etc). You could still get duplicate
> violations due to concurrent activity in READ COMMITTED mode, but not
> at higher isolation levels thanks to Thomas Munro's work there. In
> this world, ON CONFLICT and MERGE are fairly distinct things.
FWIW, and as an outsider, having looked at MERGE docs from other
RDBMSes, I have to agree that the PG UPSERT (ON CONFLICT .. DO) and
MERGE are rather different beasts.
In particular, I suspect all UPSERT statements can be mapped onto
equivalent MERGE statements, but not all MERGE statements can be mapped
onto UPSERTs.
The reason is that UPSERT depends on UNIQUE constraints, whereas MERGE
uses a generic join condition that need not even refer to any INDEXes,
let alone UNIQUE ones.
Perhaps PG's UPSERT can be generalized to create a temporary UNIQUE
constraint on the specified in the conflict_target portion of the
statement, increasing the number of MERGE statements that could be
mapped onto UPSERT. But even then, that would still be a UNIQUE
constraint, whereas MERGE does not even imply such a thing.
Now, a subset of MERGE (those using equijoins in the ON condition) can
be mapped onto UPSERT provided either a suitable UNIQUE index exists (or
that PG notionally creates a temporary UNIQUE constraint for the purpose
of evaluating the UPSERT). This approach would NOT preclude a more
complete subsequent implementation of MERGE. But I wonder if that's
worthwhile given that a proper and complete implementation of MERGE is
probably very desirable.
On a tangentially related note, I've long wanted to have an RDBMS-
independent SQL parser for the purpose of implementing external query-
rewriting (and external optimizers), syntax highlighting, and so on.
Having an external / plug-in method for rewriting unsupported SQL as a
way of bridging functionality gaps (like lack of MERGE support) would be
very nice. PG does have a way to expose its AST... It might be a good
idea to start by implementing unsupported SQL features in such a way
that they parse and can produce an AST along with a syntax/unsupported
error -- then one might rewrite the parsed AST, generate appropriate
SQL, and execute that.
Nico
--
--
Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org)
To make changes to your subscription:
http://www.postgresql.org/mailpref/pgsql-hackers
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Reply to all the recipients using the --to and --cc options:
reply via email
To: pgsql-hackers@postgresql.org
Cc: nico@cryptonector.com, pg@bowt.ie, simon@2ndquadrant.com, robertmhaas@gmail.com, michael.paquier@gmail.com
Subject: Re: MERGE SQL Statement for PG11
In-Reply-To: <20171030183649.GM4496@localhost>
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
This inbox is served by DDX for PostgreSQL; see mirroring instructions
for how to clone and mirror all data and code used for this inbox