| From: | David Rowley <dgrowleyml(at)gmail(dot)com> |
|---|---|
| To: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
| Cc: | Jeevan Chalke <jeevan(dot)chalke(at)enterprisedb(dot)com>, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org> |
| Subject: | Re: unrecognized node type while displaying a Path due to dangling pointer |
| Date: | 2023-07-12 00:16:18 |
| Message-ID: | CAApHDvqo24FeYaXbB1L30ZXFzi+t5Pf3=ahqPQmcRMmy0kmS=w@mail.gmail.com |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
On Wed, 12 Jul 2023 at 08:46, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> A low-cost fix perhaps could be to unlink the lower rel's whole
> path list (set input_rel->pathlist = NIL, also zero the related
> fields such as cheapest_path) once we've finished selecting the
> paths we want for the upper rel. That's not great for debuggability
> either, but maybe it's the most appropriate compromise.
I've not taken the time to fully understand this, but from reading the
thread, I'm not immediately understanding why we can't just shallow
copy the Path from the other RelOptInfo and replace the parent before
using it in the upper RelOptInfo. Can you explain?
David
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Andres Freund | 2023-07-12 00:36:47 | Re: Support to define custom wait events for extensions |
| Previous Message | Peter Smith | 2023-07-12 00:04:09 | Re: [Patch] Use *other* indexes on the subscriber when REPLICA IDENTITY is FULL |