|From:||Ashutosh Bapat <ashutosh(dot)bapat(at)enterprisedb(dot)com>|
|To:||Amit Langote <Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp>|
|Cc:||Pg Hackers <pgsql-hackers(at)postgresql(dot)org>|
|Subject:||Re: dropping partitioned tables without CASCADE|
|Views:||Raw Message | Whole Thread | Download mbox|
On Wed, Feb 22, 2017 at 11:11 AM, Amit Langote
> On 2017/02/22 13:46, Ashutosh Bapat wrote:
>> Looks good to me. In the attached patch I have added a comment
>> explaining the reason to make partition tables "Auto" dependent upon
>> the corresponding partitioned tables.
> Good call.
> + /*
> + * Unlike inheritance children, partition tables are expected to be dropped
> + * when the parent partitioned table gets dropped.
> + */
> Hmm. Partitions *are* inheritance children, so we perhaps don't need the
> part before the comma. Also, adding "automatically" somewhere in there
> would be nice.
> Or, one could just write: /* add an auto dependency for partitions */
I changed it in the attached patch to
+ * Partition tables are expected to be dropped when the parent partitioned
+ * table gets dropped.
>> In the tests we are firing commands to drop partitioned table, but are
>> not checking whether those tables or the partitions are getting
>> dropped or not. Except for drop_if_exists.sql, I did not find that we
>> really check this. Should we try a query on pg_class to ensure that
>> the tables get really dropped?
> I don't see why this patch should do it, if dependency.sql itself does
> not? I mean dropping AUTO dependent objects is one of the contracts of
> dependency.c, so perhaps it would make sense to query pg_class in
> dependency.sql to check if AUTO dependencies work correctly.
Hmm, I agree.
The Postgres Database Company
|Next Message||Dave Page||2017-02-22 12:47:33||Re: pg_monitor role|
|Previous Message||Bernd Helmle||2017-02-22 12:23:52||Re: [patch] reorder tablespaces in basebackup tar stream for backup_label|