Hi Michael,
On 2/24/20 7:26 AM, Michael Paquier wrote:
> On Sun, Feb 23, 2020 at 04:08:58PM +0900, Michael Paquier wrote:
>> Good idea. Let's do things as you suggest.
>
> Applied and back-patched this one down to 11.
FWIW, we took a slightly narrower approach to this issue in the
pgBackRest patch (attached).
I don't have an issue with the prefix approach since it works and the
Postgres project is very likely to catch it if there is a change in
behavior.
For third-party projects, though, it might pay to be more conservative
in case the behavior changes in the future, i.e.
pg_internal.init[something] (but not pg_internal\.init[0-9]+) becomes valid.
Regards,
--
-David
david@pgmasters.net