MIME-Version: 1.0
From: Brian McMahon <brianmcmahon135@gmail.com>
Date: Tue, 30 Jul 2024 08:53:55 -0700
Message-ID: 
 <CA+0iD35fvs6cXybmj+FSY_QWpir_c60gGzoi+GX+jTLHEf0Z-A@mail.gmail.com>
Subject: Support for on conflict ignore returning
To: pgsql-novice@lists.postgresql.org
Content-Type: multipart/alternative; boundary="000000000000307ad0061e78fdbe"
Archived-At: 
 <https://www.postgresql.org/message-id/CA%2B0iD35fvs6cXybmj%2BFSY_QWpir_c60gGzoi%2BGX%2BjTLHEf0Z-A%40mail.gmail.com>
Precedence: bulk

--000000000000307ad0061e78fdbe
Content-Type: text/plain; charset="UTF-8"

Hi!

I have a question about support / if there's plans to support a feature, I
thought it would be best to send it to the "novice" mailing list because it
might be a naive question that's been answered before.

Is there a recommended way to achieve a ON CONFLICT DO NOTHING RETURNING
<column> so that conflicting columns still have information returned?
Currently it seems like the only simple way to achieve this is with a NOOP
update, so instead doing something like ON CONFLICT UPDATE SET name =
EXCLUDED.name RETURNING <column>.

As I understand it, this isn't desirable as there's various potentially
undesirable side effects since it's not treated as a NOOP, it's treated as
an update so triggers and all sorts of internal postgres logic are invoked
(which may or may not be fine for certain applications, but it's not
desirable).

Are there current features that support this functionality, or plans to
support something like this in the future?

I should probably add my use case, I need to insert a bunch of rows into
the database to ensure they exist, then for all those rows I need to update
corresponding tables that reference them with more data. If I can't make a
query that returns the primary key, then I either need to make a subsequent
query to pull all the rows from the database that weren't inserted to get
the primary keys, or I need to use the unique fields that caused the
conflict in the next query to update the tables that reference the table I
updated, which makes things a little messier (I'd rather just work with the
primary keys when possible).

Thanks in advance.

-- 
Sincerely,
Brian McMahon

--000000000000307ad0061e78fdbe
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable

<div dir=3D"ltr">Hi!<div><br></div><div>I have a question about support / i=
f there&#39;s plans to support a feature, I thought it would be best to sen=
d it to the &quot;novice&quot; mailing list because it might be a naive que=
stion that&#39;s been answered before.</div><div><br></div><div>Is there a =
recommended way to achieve a ON CONFLICT DO NOTHING RETURNING &lt;column&gt=
; so that conflicting columns still have information returned? Currently it=
 seems like the only simple way to achieve this is with=C2=A0a NOOP update,=
 so instead doing something like ON CONFLICT UPDATE SET name =3D EXCLUDED.n=
ame RETURNING &lt;column&gt;.</div><div><br></div><div>As I understand it, =
this isn&#39;t desirable as there&#39;s various potentially undesirable sid=
e effects since it&#39;s not treated as a NOOP, it&#39;s treated as an upda=
te so triggers and all sorts of internal postgres logic are invoked (which =
may or may not be fine for certain applications, but it&#39;s not desirable=
).</div><div><div><br></div><div>Are there current features that support=C2=
=A0this functionality, or plans to support something like this in the futur=
e?</div><div><br></div><div>I should probably add my use case, I need to in=
sert a bunch of rows into the database to ensure they exist, then for all t=
hose rows I need to update corresponding tables that reference them with mo=
re data. If I can&#39;t make a query that returns the primary key, then I e=
ither need to make a subsequent query to pull all the rows from the databas=
e that weren&#39;t inserted to get the primary keys, or I need to use the u=
nique fields that caused the conflict in the next query to update the table=
s that reference the table I updated, which makes things a little messier (=
I&#39;d rather just work with the primary keys when possible).</div><div><b=
r></div><div>Thanks in advance.</div><div><br></div><span class=3D"gmail_si=
gnature_prefix">-- </span><br><div dir=3D"ltr" class=3D"gmail_signature" da=
ta-smartmail=3D"gmail_signature"><div dir=3D"ltr">Sincerely,<div>Brian McMa=
hon</div></div></div></div></div>

--000000000000307ad0061e78fdbe--