Message-ID: <ebfa8b0a-5858-4167-884c-499e70fce2cd@iki.fi>
Date: Tue, 9 Dec 2025 22:04:43 +0200
MIME-Version: 1.0
User-Agent: Mozilla Thunderbird
Subject: Re: pgsql: injection_points: Remove portions related to custom
 pgstats
To: Tom Lane <tgl@sss.pgh.pa.us>
Cc: Michael Paquier <michael@paquier.xyz>,
 Andrew Dunstan <andrew@dunslane.net>, pgsql-committers@lists.postgresql.org,
 Sami Imseih <samimseih@gmail.com>
References: <E1vSSCn-003N0Z-2i@gemulon.postgresql.org>
 <2899652.1765167313@sss.pgh.pa.us> <aTZZzFhQTtUKWGLm@paquier.xyz>
 <2904427.1765169854@sss.pgh.pa.us> <aTaPKpYufxut5Zs1@paquier.xyz>
 <d92f0eca-bb14-4b8c-9fb0-3b79635e1a64@iki.fi>
 <251242.1765302019@sss.pgh.pa.us>
Content-Language: en-US
From: Heikki Linnakangas <hlinnaka@iki.fi>
In-Reply-To: <251242.1765302019@sss.pgh.pa.us>
Content-Type: text/plain; charset=UTF-8; format=flowed
Content-Transfer-Encoding: 7bit
Archived-At: 
 <https://www.postgresql.org/message-id/ebfa8b0a-5858-4167-884c-499e70fce2cd%40iki.fi>
Precedence: bulk

On 09/12/2025 19:40, Tom Lane wrote:
> Perhaps another alternative is to add "DROP EXTENSION
> injection_points" at the end of the test scripts in the
> problematic modules.

It's a little tricky if a module has more than one test that uses the 
extension. Namely src/test/modules/nbtree which I added recently has two 
test scripts, nbtree_incomplete_splits.sql and 
nbtree_half_dead_pages.sql, and both of them do "create extension if not 
exists injection_points;". They run concurrently, so they race on which 
one creates the extension first. We can't do the same for DROP EXTENSION 
at the end.

Maybe that's not a great design in the first place, though. Perhaps that 
module should use a schedule file:

# create injection_points extension before the actual tests
test: test_setup

test: nbtree_incomplete_splits nbtree_half_dead_pages

# drop injection_points extension because it cannot be pg_upgraded.
test: test_clean

- Heikki