From: | Robert Haas <robertmhaas(at)gmail(dot)com> |
---|---|
To: | Ashutosh Bapat <ashutosh(dot)bapat(at)enterprisedb(dot)com> |
Cc: | pgsql-hackers <pgsql-hackers(at)postgresql(dot)org>, Andreas Seltenreich <seltenreich(at)gmx(dot)de> |
Subject: | Re: Re: [COMMITTERS] pgsql: Expand partitioned table RTEs level by level, without flattening |
Date: | 2017-09-20 14:21:53 |
Message-ID: | CA+TgmoYLrRrOjbrTZrtxvzXKPf0v2JH3SqiFj7tR9tWvvkzTxQ@mail.gmail.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-committers pgsql-hackers |
On Mon, Sep 18, 2017 at 3:30 AM, Ashutosh Bapat
<ashutosh(dot)bapat(at)enterprisedb(dot)com> wrote:
>> testing with sqlsmith shows that the following assertion in this commit
>> doesn't hold:
>>
>> TRAP: FailedAssertion("!(((brel)->reloptkind == RELOPT_BASEREL ||
>> (brel)->reloptkind == RELOPT_OTHER_MEMBER_REL))", File: "initsplan.c", Line:
>> 647)
>>
>> One of the simpler queries that triggers it for me:
>>
>> select from information_schema.user_mapping_options;
>
> Thanks Andreas for the report. Sorry for the assertion failure.
>
> PFA patch to fix the assertion failure.
>
> The assertion assumed that all relations in simple_rel_array[] were
> either "base" relations or "other" relations. This isn't true. The
> array can contain "dead" relations as well. I have removed the
> assertion and instead fixed the code to skip anything which is not
> "base" or "other member rel".
>
> I have also added a test to cover dead relations and lateral
> references in join.sql.
Committed. Thanks.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company
From | Date | Subject | |
---|---|---|---|
Next Message | Tom Lane | 2017-09-20 15:10:54 | pgsql: Fix erroneous documentation about noise word GROUP. |
Previous Message | Robert Haas | 2017-09-20 14:21:28 | pgsql: Fix create_lateral_join_info to handle dead relations properly. |
From | Date | Subject | |
---|---|---|---|
Next Message | Rajkumar Raghuwanshi | 2017-09-20 14:25:53 | Re: Partition-wise join for join between (declaratively) partitioned tables |
Previous Message | Alexander Korotkov | 2017-09-20 14:09:38 | Re: Allow GiST opcalsses without compress\decompres functions |