Re: BUG #18440: Query does not prune partitions correctly or use index when prepared statements are used

On Tue, Apr 16, 2024 at 11:56 AM PG Bug reporting form <
noreply(at)postgresql(dot)org> wrote:

> The following bug has been logged on the website:
>
> Bug reference: 18440
> Logged by: Damir Ciganović-Janković
> Email address: damir(dot)ciganovic(dot)jankovic(at)gmail(dot)com
> PostgreSQL version: 15.6
> Operating system: Linux
> Description:
>
> Note, same bug from
>
> https://www.postgresql.org/message-id/17484-716a1fcbcb0e379b%40postgresql.org
> , I was instructed to check on Postgres 14, but the issue is still present
> in 15.6 and 16.2 versions.

Like with the other report this isn't a bug report. It is a symptom of
difficulty in explaining how all these complex features interact with each
other and thus users having the wrong expectations.

Compare the plans for a circa v13 execution and a circa v14 execution to
see what was gained.

https://dbfiddle.uk/tcwTciT9

In short, expecting even a custom plan of a prepared statement to match an
unprepared statement is incorrect. There is still considerable saved
worked done without knowledge of parameters that has to be ignored, not
discarded.

David J.