Message-ID: <548a47bc-87ae-b3df-c6a2-60b9966f808b@gmail.com>
Date: Wed, 1 May 2024 16:00:00 +0300
MIME-Version: 1.0
User-Agent: Mozilla/5.0 (X11; Linux x86_64; rv:102.0) Gecko/20100101
 Thunderbird/102.4.2
Subject: Re: BUG #18449: Altering column type fails when an SQL routine
 depends on the column
Content-Language: en-US
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: pgsql-bugs@lists.postgresql.org,
 Peter Eisentraut <peter.eisentraut@enterprisedb.com>
References: <18449-f8248467aaa294d5@postgresql.org>
 <403599.1714268901@sss.pgh.pa.us>
From: Alexander Lakhin <exclusion@gmail.com>
In-Reply-To: <403599.1714268901@sss.pgh.pa.us>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/548a47bc-87ae-b3df-c6a2-60b9966f808b%40gmail.com>
Precedence: bulk

Hello Tom,

28.04.2024 04:48, Tom Lane wrote:
> Thanks for the report.  It looks like most of the other hard cases
> in RememberAllDependentForRebuilding just error out with code
> along the lines of
>
>                      ereport(ERROR,
>                              (errcode(ERRCODE_FEATURE_NOT_SUPPORTED),
>                               errmsg("cannot alter type of a column used by a view or rule"),
>
> I'm inclined to do likewise for functions.

I've discovered one more case, presumably as hard as the other ones:
CREATE TABLE t(a int);
CREATE PUBLICATION p FOR TABLE t WHERE (a > 0);

ALTER TABLE t ALTER COLUMN a TYPE bigint;
fails with:
ERROR:  unexpected object depending on column: publication of table t in publication p

Reproduced on REL_15_STABLE (52e4f0cd4) .. master.

> We could imagine trying
> to re-parse the function definition against the new column type,
> but there are way too many ways for that to go wrong.  Just for
> starters, there are possibly-security-grade hazards if the current
> search_path isn't what it was when the function was created.  There's
> no guarantee that we'd succeed anyway, eg the new column type might
> not work for some function or operator that's applied to it, and if
> not the resulting error message would likely be very surprising.

It looks like all these considerations apply to expressions defining row
filters for publications...

Best regards,
Alexander