Clarify how triggers relate to transactions
The following documentation comment has been logged on the website:
Page: https://www.postgresql.org/docs/13/sql-createtrigger.html
Description:
https://www.postgresql.org/docs/current/sql-createtrigger.html mentions the
word "transaction" only once, in reference specifically to constraint
triggers: "They can be fired either at the end of the statement causing the
triggering event, or at the end of the containing transaction; in the latter
case they are said to be deferred."
If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the same transaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."
On Tue, 2021-04-27 at 14:26 +0000, PG Doc comments form wrote:
The following documentation comment has been logged on the website:
Page: https://www.postgresql.org/docs/13/sql-createtrigger.html
Description:https://www.postgresql.org/docs/current/sql-createtrigger.html mentions the
word "transaction" only once, in reference specifically to constraint
triggers: "They can be fired either at the end of the statement causing the
triggering event, or at the end of the containing transaction; in the latter
case they are said to be deferred."If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the same transaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."
Good idea in principle, but I'd put that information on
https://www.postgresql.org/docs/current/trigger-definition.html
Yours,
Laurenz Albe
Gotcha. Where would I go to make the PR?
On Wed, Apr 28, 2021, 7:24 AM Laurenz Albe <laurenz.albe@cybertec.at> wrote:
Show quoted text
On Tue, 2021-04-27 at 14:26 +0000, PG Doc comments form wrote:
The following documentation comment has been logged on the website:
Page: https://www.postgresql.org/docs/13/sql-createtrigger.html
Description:https://www.postgresql.org/docs/current/sql-createtrigger.html mentions
the
word "transaction" only once, in reference specifically to constraint
triggers: "They can be fired either at the end of the statement causingthe
triggering event, or at the end of the containing transaction; in the
latter
case they are said to be deferred."
If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the sametransaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."Good idea in principle, but I'd put that information on
https://www.postgresql.org/docs/current/trigger-definition.htmlYours,
Laurenz Albe
On Wed, 2021-04-28 at 13:18 -0400, Nathan Long wrote:
Gotcha. Where would I go to make the PR?
You'd create a patch against Git master and send it to this
mailing list or pgsql-hackers. If you don't want it to fall
between the cracks, register in the next commitfest where it
can undergo peer review.
Yours,
Laurenz Albe
On Wed, Apr 28, 2021, at 2:18 PM, Nathan Long wrote:
Gotcha. Where would I go to make the PR?
There is no such PR feature; we don't use GitHub despite of having a mirror
there. As Laurenz said you should create a patch (using your preferred git
command) and attach to this thread. If you prefer, you can also send the patch
to pgsql-hackers ML (add the link to this thread). The next step is to register
your patch to the next commitfest [1]https://commitfest.postgresql.org/33/ so we don't lose track of it. For a
complete reference about submitting a patch, take a look at [2]https://wiki.postgresql.org/wiki/Submitting_a_Patch.
[1]: https://commitfest.postgresql.org/33/
[2]: https://wiki.postgresql.org/wiki/Submitting_a_Patch
Regards,
--
Euler Taveira
EDB https://www.enterprisedb.com/
On Wed, 2021-04-28 at 13:24 +0200, Laurenz Albe wrote:
On Tue, 2021-04-27 at 14:26 +0000, PG Doc comments form wrote:
https://www.postgresql.org/docs/current/sql-createtrigger.html mentions the
word "transaction" only once, in reference specifically to constraint
triggers: "They can be fired either at the end of the statement causing the
triggering event, or at the end of the containing transaction; in the latter
case they are said to be deferred."If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the same transaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."Good idea in principle, but I'd put that information on
https://www.postgresql.org/docs/current/trigger-definition.html
Here is a proposed patch for this.
Yours,
Laurenz Albe
Attachments:
0001-Document-how-triggers-interact-with-transactions.patchtext/x-patch; charset=UTF-8; name=0001-Document-how-triggers-interact-with-transactions.patchDownload
From 393cdf38d812b337b51e3f5ca2af6ca052ed1b17 Mon Sep 17 00:00:00 2001
From: Laurenz Albe <laurenz.albe@cybertec.at>
Date: Wed, 5 May 2021 11:53:38 +0200
Subject: [PATCH] Document how triggers interact with transactions
Clarify that triggers are always executed as part of the same
transaction as the triggering statement, with all consequences.
Add an index term for constraint triggers and mention them
in the documentation of trigger behavior.
---
doc/src/sgml/ref/create_trigger.sgml | 4 ++++
doc/src/sgml/trigger.sgml | 7 ++++++-
2 files changed, 10 insertions(+), 1 deletion(-)
diff --git a/doc/src/sgml/ref/create_trigger.sgml b/doc/src/sgml/ref/create_trigger.sgml
index 561af989a4..e46b9e79db 100644
--- a/doc/src/sgml/ref/create_trigger.sgml
+++ b/doc/src/sgml/ref/create_trigger.sgml
@@ -175,6 +175,10 @@ CREATE [ OR REPLACE ] [ CONSTRAINT ] TRIGGER <replaceable class="parameter">name
</para>
<para>
+ <indexterm>
+ <primary>trigger</primary>
+ <secondary>constraint trigger</secondary>
+ </indexterm>
When the <literal>CONSTRAINT</literal> option is specified, this command creates a
<firstterm>constraint trigger</firstterm>. This is the same as a regular trigger
except that the timing of the trigger firing can be adjusted using
diff --git a/doc/src/sgml/trigger.sgml b/doc/src/sgml/trigger.sgml
index 4a0e74652f..0c8cc8d0fe 100644
--- a/doc/src/sgml/trigger.sgml
+++ b/doc/src/sgml/trigger.sgml
@@ -47,7 +47,9 @@
the <command>UPDATE</command> statement. Triggers can also fire
for <command>TRUNCATE</command> statements. If a trigger event occurs,
the trigger's function is called at the appropriate time to handle the
- event.
+ event. All triggers are executed as part of the same transaction as
+ the statement that triggered them, so if either the statement or the
+ trigger causes an error, the effects of both will be rolled back.
</para>
<para>
@@ -115,6 +117,9 @@
<literal>BEFORE</literal> triggers fire immediately before a particular row is
operated on, while row-level <literal>AFTER</literal> triggers fire at the end of
the statement (but before any statement-level <literal>AFTER</literal> triggers).
+ The execution of <literal>AFTER</literal> triggers can also be deferred to the
+ end of the transaction if they were defined as
+ <firstterm>constraint triggers</firstterm>.
These types of triggers may only be defined on tables and
foreign tables, not views.
<literal>INSTEAD OF</literal> triggers may only be
--
2.26.3
On Wed, 2021-05-05 at 11:55 +0200, Laurenz Albe wrote:
On Wed, 2021-04-28 at 13:24 +0200, Laurenz Albe wrote:
On Tue, 2021-04-27 at 14:26 +0000, PG Doc comments form wrote:
https://www.postgresql.org/docs/current/sql-createtrigger.html mentions the
word "transaction" only once, in reference specifically to constraint
triggers: "They can be fired either at the end of the statement causing the
triggering event, or at the end of the containing transaction; in the latter
case they are said to be deferred."If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the same transaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."Good idea in principle, but I'd put that information on
https://www.postgresql.org/docs/current/trigger-definition.htmlHere is a proposed patch for this.
Replying to -hackers for the commitfest app.
Yours,
Laurenz Albe
Laurenz Albe <laurenz.albe@cybertec.at> writes:
On Wed, 2021-04-28 at 13:24 +0200, Laurenz Albe wrote:
On Tue, 2021-04-27 at 14:26 +0000, PG Doc comments form wrote:
If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the same transaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."
Good idea in principle, but I'd put that information on
https://www.postgresql.org/docs/current/trigger-definition.html
Here is a proposed patch for this.
I think this is a good idea, but I felt like you'd added the extra
sentences in not-terribly-well-chosen places. For instance, your
first addition in trigger.sgml is adding to a para that talks about
triggers for tables, while the next para talks about triggers for
views. So it seems unclear whether the statement is meant to apply
to view triggers too.
I think it'd work out best to make this a separate para after the
one that defines before/after/instead-of triggers. How do you
like the attached?
regards, tom lane
Attachments:
0001-Document-how-triggers-interact-with-transactions-2.patchtext/x-diff; charset=us-ascii; name*0=0001-Document-how-triggers-interact-with-transactions-2.pat; name*1=chDownload
diff --git a/doc/src/sgml/ref/create_trigger.sgml b/doc/src/sgml/ref/create_trigger.sgml
index 3f4b5acc7b..e4afa1c01d 100644
--- a/doc/src/sgml/ref/create_trigger.sgml
+++ b/doc/src/sgml/ref/create_trigger.sgml
@@ -175,6 +175,10 @@ CREATE [ OR REPLACE ] [ CONSTRAINT ] TRIGGER <replaceable class="parameter">name
</para>
<para>
+ <indexterm>
+ <primary>trigger</primary>
+ <secondary>constraint trigger</secondary>
+ </indexterm>
When the <literal>CONSTRAINT</literal> option is specified, this command creates a
<firstterm>constraint trigger</firstterm>. This is the same as a regular trigger
except that the timing of the trigger firing can be adjusted using
diff --git a/doc/src/sgml/trigger.sgml b/doc/src/sgml/trigger.sgml
index f1a845f756..7e2654493b 100644
--- a/doc/src/sgml/trigger.sgml
+++ b/doc/src/sgml/trigger.sgml
@@ -122,6 +122,15 @@
row in the view is identified as needing to be operated on.
</para>
+ <para>
+ The execution of an <literal>AFTER</literal> trigger can be deferred
+ to the end of the transaction, rather than the end of the statement,
+ if it was defined as a <firstterm>constraint trigger</firstterm>.
+ In all cases, a trigger is executed as part of the same transaction as
+ the statement that triggered it, so if either the statement or the
+ trigger causes an error, the effects of both will be rolled back.
+ </para>
+
<para>
A statement that targets a parent table in an inheritance or partitioning
hierarchy does not cause the statement-level triggers of affected child
On Fri, 2021-07-30 at 16:20 -0400, Tom Lane wrote:
Laurenz Albe <laurenz.albe@cybertec.at> writes:
On Wed, 2021-04-28 at 13:24 +0200, Laurenz Albe wrote:
On Tue, 2021-04-27 at 14:26 +0000, PG Doc comments form wrote:
If I understand correctly, it would be helpful to add this sentence or a
corrected version of it: "Triggers always execute in the same transaction as
the triggering event, and if a trigger fails, the transaction is rolled
back."Here is a proposed patch for this.
I think this is a good idea, but I felt like you'd added the extra
sentences in not-terribly-well-chosen places. For instance, your
first addition in trigger.sgml is adding to a para that talks about
triggers for tables, while the next para talks about triggers for
views. So it seems unclear whether the statement is meant to apply
to view triggers too.I think it'd work out best to make this a separate para after the
one that defines before/after/instead-of triggers. How do you
like the attached?
That is better, and I like your patch. Thanks!
Keeping paragraphs short is a good thing.
Yours,
Laurenz Albe
Laurenz Albe <laurenz.albe@cybertec.at> writes:
On Fri, 2021-07-30 at 16:20 -0400, Tom Lane wrote:
I think it'd work out best to make this a separate para after the
one that defines before/after/instead-of triggers. How do you
like the attached?
That is better, and I like your patch. Thanks!
Keeping paragraphs short is a good thing.
Pushed like that, then.
regards, tom lane