Re: pg_get_multixact_members not documented - Mailing list pgsql-hackers

From Nathan Bossart
Subject Re: pg_get_multixact_members not documented
Date
Msg-id aD4DoMW1L5xxIh3K@nathan
Whole thread Raw
In response to Re: pg_get_multixact_members not documented  (Nathan Bossart <nathandbossart@gmail.com>)
List pgsql-hackers
On Mon, Jun 02, 2025 at 12:46:51PM -0500, Sami Imseih wrote:
> v1-0001 is the documentation only patch. I improved upon the description
> suggested in [0]

Your patch adds an entry to the "Transaction ID and Snapshot Information
Functions" table, while Álvaro's introduced a new "Multixact Functions"
table.  His also added a note to maintenance.sgml.  Any reason for the
differences?

> A simple test will be a regress/sql which ensure the XID and lock mode
> of a transaction using a savepoint, something like the below. To do anything
> fancier with concurrency, we will need an isolation test.
> 
> ```
> drop table if exists t;
> create table t (v int);  insert into t values (1);
> begin;
> select from t for update ;
> savepoint s1;
> update t set v = v;
> select pg_get_multixact_members(a.relminmxid), a.relminmxid from
> (select relminmxid from pg_class where relname = 't') a;
> commit;
> ```

That seems reasonable to me.

-- 
nathan



pgsql-hackers by date:

Previous
From: Alena Rybakina
Date:
Subject: Re: Vacuum statistics
Next
From: Peter Geoghegan
Date:
Subject: Re: Correcting freeze conflict horizon calculation