Message-ID: <3b0e067d-9a78-46f7-8027-99c19e7cde44@illuminatedcomputing.com>
Date: Wed, 6 Nov 2024 17:03:09 -0800
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: SQL:2011 application time
To: Sam Gabrielsson <sam@movsom.se>
Cc: Peter Eisentraut <peter@eisentraut.org>,
 jian he <jian.universality@gmail.com>,
 Isaac Morland <isaac.morland@gmail.com>, Robert Haas
 <robertmhaas@gmail.com>, Jeff Davis <pgsql@j-davis.com>,
 PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
References: 
 <CA+renyUApHgSZF9-nd-a0+OPGharLQLO=mDHcY4_qQ0+noCUVg@mail.gmail.com>
 <30dc3561-d7ab-42be-a1c5-096643423590@illuminatedcomputing.com>
 <b824fc91-f534-4096-8d94-de8e21b06915@illuminatedcomputing.com>
 <73d09e24-b6b9-4c89-8c0a-8842ca0f81f9@illuminatedcomputing.com>
 <c002230c-8879-4b2c-ad05-585d6572347c@illuminatedcomputing.com>
 <CACJufxESqArbu=R8m3a+ZF-Yia26KTGpEnyqz+8d5BXb1o=c=w@mail.gmail.com>
 <CACJufxECGaLjMcN8TJbX1fGrMFKs6qu+6RiZm98=sLcpiVpu2g@mail.gmail.com>
 <081aa43c-3a97-4bd5-a498-ba02e0a7570d@illuminatedcomputing.com>
 <54390149-11d6-4f52-bbe8-f69832ceb3bd@eisentraut.org>
 <eedaaca9-d615-4934-ab00-39ec376d9b37@eisentraut.org>
 <CA+renyUFC13F0tYKxEENZtWA0YVuS5Tv+ZQkEkAwuDO1-Xke-A@mail.gmail.com>
 <9b7cb814-c78c-489f-85b2-32be4398647f@illuminatedcomputing.com>
 <95c6d49b-7d60-4ad0-820c-e00ae1b2a3d4@illuminatedcomputing.com>
 <57ea0668-5205-426e-b934-efc89f2186c2@illuminatedcomputing.com>
 <e23b9d86e07301e8a38a3d0617e3719c@movsom.se>
Content-Language: en-US
From: Paul Jungwirth <pj@illuminatedcomputing.com>
In-Reply-To: <e23b9d86e07301e8a38a3d0617e3719c@movsom.se>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 8bit
Archived-At: 
 <https://www.postgresql.org/message-id/3b0e067d-9a78-46f7-8027-99c19e7cde44%40illuminatedcomputing.com>
Precedence: bulk

On 11/4/24 13:16, Sam Gabrielsson wrote:
> Foreign key violation errors are incorrectly raised in a few cases for a temporal foreign key with 
> default ON UPDATE NO ACTION. Test is based on the commited v39 patches (used a snapshot version of 
> PG18 devel available from PGDG).

Thank you for the report! I confirmed that this is a problem. In ri_restrict we fail if any fk 
records still match the being-changed pk, but for temporal if you're merely shrinking the pk range, 
fk references could still wind up being valid (if you're only shrinking it a little). So we need to 
do more work.

> In the temporal NO ACTION case something similar to this (though with appropriate locks) could 
> perhaps be tested in ri_restrict (when ri_Check_Pk_Match returns false):
> 
>    SELECT 1
>    FROM (SELECT range_agg(pkperiodatt) AS r
>        FROM <pktable>
>        WHERE pkatt1 = $1 [AND ...]
>        AND pkperiodatt && $n) AS pktable,
>      (SELECT fkperiodatt AS r
>        FROM <fktable>
>        WHERE fkatt1 = $1 [AND ...]
>        AND fkperiodatt && $n) AS fktable
>    WHERE NOT fktable.r <@ pktable.r

This solution looks like it will work to me. Basically: find FKs that still match the PK, but only 
fail if they are no longer covered.

IIRC for RESTRICT it is *correct* to reject the change, so we would want to keep the old SQL there, 
and only update it for NOACTION.

I'll work on a fix and submit another set of patches.

Yours,

-- 
Paul              ~{:-)
pj@illuminatedcomputing.com