Horton

Re: Early WIP/PoC for inlining CTEs

Andres Freund <andres@anarazel.de>

View thread
From: Andres Freund <andres@anarazel.de>
To: Thomas Munro <thomas.munro@enterprisedb.com>
Cc: David Fetter <david@fetter.org>, Andrew Gierth <andrew@tao11.riddles.org.uk>, Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2018-08-08T05:02:11Z
Lists: pgsql-hackers

Commits

Same data as JSON: GET /api/v1/messages/:b64id/commits the thread's linked commits as JSON, with link sources. API reference →

  1. Prevent inlining of multiply-referenced CTEs with outer recursive refs.

  2. Allow user control of CTE materialization, and change the default behavior.

  3. Split QTW_EXAMINE_RTES flag into QTW_EXAMINE_RTES_BEFORE/_AFTER.

  4. document when PREPARE uses generic plans

Hi,

On 2018-08-08 16:55:22 +1200, Thomas Munro wrote:
> On Fri, Jul 27, 2018 at 8:10 PM, David Fetter <david@fetter.org> wrote:
> > On Fri, Jul 27, 2018 at 02:55:26PM +1200, Thomas Munro wrote:
> >> On Thu, Jul 26, 2018 at 7:14 AM, David Fetter <david@fetter.org> wrote:
> >> > Please find attached the next version, which passes 'make check'.
> >>
> >> ... but not 'make check-world' (contrib/postgres_fdw's EXPLAIN is different).
> >
> > Please find attached a patch that does.
> >
> > It doesn't always pass make installcheck-world, but I need to sleep
> > rather than investigate that at the moment.
> 
> One observation I wanted to share: CTE scans inhibit parallelism today
> (something we might eventually want to fix with shared tuplestores).
> This patch therefore allows parallelism in some WITH queries, which
> seems like a very valuable thing.

Might be interesting to see how big a difference it makes for
TPC-DS. Currently the results are bad (as in many queries don't finish
in a relevant time) because it uses CTEs so widely, and there's often
predicates outside the CTE that could be pushed down.

- Andres