MIME-Version: 1.0
References: <CACJufxHJAr2FjbeB6ghg_-N5dxX5JVnjKSLOUxOyt4TeaAWQkg@mail.gmail.com>
 <CACJufxHQ8tDG09mXW=8xfEK2_h+By1zLgrFd0EcshHp-TDGiGw@mail.gmail.com>
 <CACJufxFu7Y4FhVkaKT2Kaj8ym2T5TcwN93cR_6h4x66iLrSZ-Q@mail.gmail.com>
 <CACJufxHMYq2ZqvOhoSKUmx+wZUo=HixjfgGKDJ6L4cdCvwh2VA@mail.gmail.com>
 <AF1AE30F-4379-462D-B321-FB1658230B5D@yandex-team.ru> <CACJufxH5YDMAQ_KXaPJaC4Zt1i2HNSNaYoskzNDRjSrE_UWjKw@mail.gmail.com>
In-Reply-To: <CACJufxH5YDMAQ_KXaPJaC4Zt1i2HNSNaYoskzNDRjSrE_UWjKw@mail.gmail.com>
From: Zsolt Parragi <zsolt.parragi@percona.com>
Date: Wed, 21 Jan 2026 12:13:04 +0000
Message-ID: <CAN4CZFM_8jL7Rf43zLa-9P2BwC+_H-oM9kYxVLGcwu83KC6faA@mail.gmail.com>
Subject: Re: CREATE TABLE LIKE INCLUDING TRIGGERS
To: jian he <jian.universality@gmail.com>
Cc: Andrey Borodin <x4mmm@yandex-team.ru>, 
	PostgreSQL-development <pgsql-hackers@postgresql.org>
Content-Type: text/plain; charset="UTF-8"
Content-Transfer-Encoding: quoted-printable
Archived-At: <https://www.postgresql.org/message-id/CAN4CZFM_8jL7Rf43zLa-9P2BwC%2B_H-oM9kYxVLGcwu83KC6faA%40mail.gmail.com>
Precedence: bulk

Hello!

I tested the patch, it works as described, but I did notice one possible is=
sue:

Shouldn't this preserve the enabled state of the triggers, or if it
doesn't, should the documentation include this limitations?

Currently the new table will always use the default
TRIGGER_FIRES_ON_ORIGIN, regardless how it is configured for the
original table.

diff --git a/src/test/regress/sql/triggers.sql
b/src/test/regress/sql/triggers.sql
index a1e9dbba8bd..24f7bfb5837 100644
--- a/src/test/regress/sql/triggers.sql
+++ b/src/test/regress/sql/triggers.sql
@@ -260,6 +260,18 @@ ON      pc.oid =3D pd.tgrelid AND pd.tgname =3D
'before_ins_stmt_trig'
 ORDER BY 1;
 COMMENT ON TRIGGER before_ins_stmt_trig ON main_table IS NULL;

+-- Test that trigger firing state is not preserved by LIKE INCLUDING TRIGG=
ERS
+ALTER TABLE main_table DISABLE TRIGGER before_ins_stmt_trig;
+CREATE TABLE main_table2 (LIKE main_table INCLUDING TRIGGERS);
+SELECT c.relname, t.tgname, t.tgenabled
+FROM pg_trigger t
+JOIN pg_class c ON t.tgrelid =3D c.oid
+WHERE t.tgname =3D 'before_ins_stmt_trig'
+  AND c.relname IN ('main_table', 'main_table2')
+ORDER BY c.relname;
+DROP TABLE main_table2;
+ALTER TABLE main_table ENABLE TRIGGER before_ins_stmt_trig;
+
 --
 -- Test case for bug with BEFORE trigger followed by AFTER trigger with WH=
EN
 --

On Wed, Jan 21, 2026 at 12:00=E2=80=AFPM jian he <jian.universality@gmail.c=
om> wrote:
>
> On Fri, Jan 2, 2026 at 5:25=E2=80=AFPM Andrey Borodin <x4mmm@yandex-team.=
ru> wrote:
> >
> > + char *trigcomment; /* comment to apply to trigger, or NULL */
> > No other Create*Stmt has a comment field. Comments seem to be handled t=
hrough separate CommentStmt creation.
> >
>
> See CreateStatsStmt.stxcomment, IndexStmt.idxcomment.
> We need CreateTrigStmt.trigcomment, because if INCLUDING COMMENTS is spec=
ified,
> CreateTrigStmt.trigcomment can be used to hold the source object's commen=
ts.
>
> > Some nitpicking about tests:
> > 1. INSTEAD OF triggers on views - The error is tested, but should also =
test that statement-level VIEW triggers work
> ok. test added.
>
> > 2. Triggers on partitioned tables - What happens when you LIKE a partit=
ioned table? Are partition triggers cloned?
> no. only the trigger on the partitioned table itself will be cloned.
> see tests ``create table parted_constr_copy (like parted_constr
> including all);``
>
> > 3. Cross-schema trigger functions - The function name reconstruction ha=
ndles schemas, but is it tested?
> >
> ok. test added.
>
> > + funcname =3D list_make2(makeString(schemaname),makeString(NameStr(pro=
cform->proname)));
> > Other NameStr() are pstrdup()'d, maybe let's pstrdup() this too?
> >
> ok.
>
> > + /* Reconstruct trigger old transition table */
> > Second instance of this comment is wrong.
> >
> ok.
>
> > + PG_KEYWORD("triggers", TRIGGERS, UNRESERVED_KEYWORD, BARE_LABEL)
> > Won't this break some user SQLs?
> it's marked as an un-reserved word, so it won't break any SQL, i think.
>
> v4-0001, v4-0002 was removed, as Robert said in [1],  now I am using
> the same ugly
> hack consistently, now code is more aligned with INCLUDING INDEXES, INCLU=
DING
> STATISTICS.
>
> pstrdup have been used in more places in generateClonedTriggerStmt.
>
>
>
> --
> jian
> https://url.avanan.click/v2/r01/___https://www.enterprisedb.com/___.YXAzO=
nBlcmNvbmE6YTpnOjE5NWMwMWNmNDE3Y2Y3Yzg3MGZkNWQyM2FkM2UxYjc4Ojc6NWM5NTo0YjU3=
NGI1Y2MzZjIyMjk5MzRiYWU0Njk1Yjk4NTJkNzg0YjMwYzMwZWFiMjIzYTQ4YjU1ZWUwZDIxZWI=
4ZWFhOnA6VDpO