| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | David Rowley <dgrowleyml(at)gmail(dot)com> |
| Cc: | Teodor Sigaev <teodor(at)sigaev(dot)ru>, Pgsql Hackers <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Bug in planner |
| Date: | 2015-04-25 20:18:25 |
| Message-ID: | 3993.1429993105@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
David Rowley <dgrowleyml(at)gmail(dot)com> writes:
> On 24 April 2015 at 21:43, Teodor Sigaev <teodor(at)sigaev(dot)ru> wrote:
>> I faced with planner error:
>> ERROR: could not find RelOptInfo for given relids
> I've done a little debugging on this too and I get the idea that in
> eqjoinsel() that min_righthand incorrectly does not have a bit set for "t3"
Yeah. The short of it seems to be that initsplan.c is too optimistic
about whether antijoins can be reordered against outer joins in their RHS.
The discussion in optimizer/README says pretty clearly that they can't
(and eqjoinsel is relying on that, per the comment therein), so I think
this is basically brain fade in translating that logic to code.
regards, tom lane
| Attachment | Content-Type | Size |
|---|---|---|
| antijoin-fix.patch | text/x-diff | 4.5 KB |
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Pavel Stehule | 2015-04-25 20:23:21 | Re: PL/pgSQL, RAISE and error context |
| Previous Message | Peter Geoghegan | 2015-04-25 20:00:02 | Re: INSERT ... ON CONFLICT syntax issues |