Re: ALTER COLUMN ... SET EXPRESSION to alter stored generated column's expression

From: Peter Eisentraut <peter(at)eisentraut(dot)org>
To: Amul Sul <sulamul(at)gmail(dot)com>
Cc: Robert Haas <robertmhaas(at)gmail(dot)com>, Ashutosh Bapat <ashutosh(dot)bapat(dot)oss(at)gmail(dot)com>, Vaibhav Dalvi <vaibhav(dot)dalvi(at)enterprisedb(dot)com>, pgsql-hackers(at)postgresql(dot)org
Subject: Re: ALTER COLUMN ... SET EXPRESSION to alter stored generated column's expression
Date: 2023-11-13 15:39:26
Message-ID: 5c844f1e-67d5-4b2f-8957-53044064c2ab@eisentraut.org
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

On 13.11.23 14:07, Amul Sul wrote:
> Also, it seems to me that the SET EXPRESSION variant should just do an
> update of the catalog table instead of a drop and re-insert.
>
> I am not sure if that is sufficient; we need to get rid of the
> dependencies of
> existing expressions on other columns and/or objects that need to be
> removed.
> The drop and re-insert does that easily.

Ok, good point.

> The documentation needs some improvements:
>
> +    ALTER [ COLUMN ] <replaceable
> class="parameter">column_name</replaceable> SET EXPRESSION <replaceable
> class="parameter">expression</replaceable> STORED
>
> If we're going to follow the Db2 syntax, there should be an "AS" after
> EXPRESSION.  And the implemented syntax requires parentheses, so they
> should appear in the documentation.
>
> Also, the keyword STORED shouldn't be there.  (The same command should
> be applicable to virtual generated columns in the future.)
>
> I have omitted "AS" intentionally, to keep syntax similar to our existing
> ALTERCOLUMN... SET DEFAULT <a_expr>.  Let me know if you want
> me to add that.

Well, my idea was to follow the Db2 syntax. Otherwise, we are adding
yet another slightly different syntax to the world. Even if we think
our idea is slightly better, it doesn't seem worth it.

> The STORED suggested by Vik[1].  I think we could skip that if there is
> no need
> to differentiate between stored and virtual columns at ALTER.

I think that suggestion was based on the idea that this would convert
non-generated columns to generated columns, but we have dropped that idea.

In response to

Responses

Browse pgsql-hackers by date

  From Date Subject
Next Message Pavel Stehule 2023-11-13 16:07:29 Re: proposal: possibility to read dumped table's name from file
Previous Message Kumar, Sachin 2023-11-13 15:06:31 Re: pg_upgrade failing for 200+ million Large Objects