Re: Broken lock management in policy.c. - Mailing list pgsql-hackers

From Peter Geoghegan
Subject Re: Broken lock management in policy.c.
Date
Msg-id CAM3SWZSXc0e3AYyyPZ7f_-weMe07kGpAMdw-2mK5WHqZGFBYAw@mail.gmail.com
Whole thread Raw
In response to Re: Broken lock management in policy.c.  (Stephen Frost <sfrost@snowman.net>)
Responses Re: Broken lock management in policy.c.  (Tom Lane <tgl@sss.pgh.pa.us>)
List pgsql-hackers
On Sun, Jan 3, 2016 at 6:00 PM, Stephen Frost <sfrost@snowman.net> wrote:
>> I'm not sure what you mean. The CREATE POLICY changes in commit
>> 43cd468cf01007f3 specifically call out the issue illustrated in my
>> example test case. There are some other changes made in that commit,
>> but they don't seem to be attempting to address this specific problem.
>> They also seem fine.
>
>
> I believe Tom's complaint was that the overall page is about CREATE POLICY,
> technically, and that the text in attempting to address the concern might be
> taken under the context of being a CREATE POLICY issue rather than a general
> RLS issue with row locking.

Really? But the problem happens as a consequence of having a
subqueries within CREATE POLICY's USING quals (as well as a
misunderstanding made by the admin about just what is possible).

-- 
Peter Geoghegan



pgsql-hackers by date:

Previous
From: Tom Lane
Date:
Subject: Re: Broken lock management in policy.c.
Next
From: Ian Barwick
Date:
Subject: Description tweak for vacuumdb