Re: Adding a pg_get_owned_sequence function? - Mailing list pgsql-hackers

From Nathan Bossart
Subject Re: Adding a pg_get_owned_sequence function?
Date
Msg-id 20240108210847.GA2796792@nathanxps13
Whole thread Raw
In response to Re: Adding a pg_get_owned_sequence function?  (Dagfinn Ilmari Mannsåker <ilmari@ilmari.org>)
Responses Re: Adding a pg_get_owned_sequence function?
List pgsql-hackers
On Mon, Jan 08, 2024 at 04:58:02PM +0000, Dagfinn Ilmari Mannsåker wrote:
> We can't make pg_get_serial_sequence(text, text) not work on identity
> columns any more, that would break existing users, and making the new
> function not work on serial columns would make it harder for people to
> migrate to it.  If you're suggesting adding two new functions,
> pg_get_identity_sequence(regclass, name) and
> pg_get_serial_sequence(regclass, name)¹, which only work on the type of
> column corresponding to the name, I don't think that's great for
> usability or migratability either.

I think these are reasonable concerns, but with this patch, we now have the
following functions:

    pg_get_identity_sequence(table regclass, column name) -> regclass
    pg_get_serial_sequence(table text, column text) -> text

If we only look at the names, it sure sounds like the first one only works
for identity columns, and the second only works for serial columns.  But
both work for identity _and_ serial.  The real differences between the two
are the parameter and return types.  Granted, this is described in the
documentation updates, but IMHO this is a kind-of bizarre state to end up
in.

-- 
Nathan Bossart
Amazon Web Services: https://aws.amazon.com



pgsql-hackers by date:

Previous
From: Robert Haas
Date:
Subject: Re: Emit fewer vacuum records by reaping removable tuples during pruning
Next
From: Nathan Bossart
Date:
Subject: Re: verify predefined LWLocks have entries in wait_event_names.txt