From: | Michael Paquier <michael(at)paquier(dot)xyz> |
---|---|
To: | Greg Rychlewski <greg(dot)rychlewski(at)gmail(dot)com> |
Cc: | pgsql-hackers(at)lists(dot)postgresql(dot)org |
Subject: | Re: DROP INDEX docs - explicit lock naming |
Date: | 2021-03-31 00:47:16 |
Message-ID: | YGPGlNbbDbvl5gHr@paquier.xyz |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
On Tue, Mar 30, 2021 at 10:33:46AM -0400, Greg Rychlewski wrote:
> While reading the documentation for DROP INDEX[1], I noticed the lock was
> described colloquially as an "exclusive" lock, which made me pause for a
> second because it's the same name as the EXCLUSIVE table lock.
>
> The attached patch explicitly states that an ACCESS EXCLUSIVE lock is
> acquired.
Indeed, this could be read as ACCESS SHARE being allowed, but that's
never the case for any of the index code paths, except if CONCURRENTLY
is involved. It is not the only place in the docs where we could do
more clarification. For instance, reindex.sgml mentions twice an
exclusive lock but that should be an access exclusive lock. To be
exact, I can spot 27 places under doc/ that could be improved. Such
changes depend on the surrounding context, of course.
--
Michael
From | Date | Subject | |
---|---|---|---|
Next Message | tsunakawa.takay@fujitsu.com | 2021-03-31 00:48:06 | RE: libpq debug log |
Previous Message | Andy Fan | 2021-03-31 00:44:53 | Re: Keep notnullattrs in RelOptInfo (Was part of UniqueKey patch series) |