Horton

Re: Declarative partitioning - another take

Amit Langote <langote_amit_f8@lab.ntt.co.jp>

View thread View on postgr.es
From: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Rajkumar Raghuwanshi <rajkumar.raghuwanshi@enterprisedb.com>, Amit Langote <amitlangote09@gmail.com>, Pg Hackers <pgsql-hackers@postgresql.org>, pgsql-hackers-owner@postgresql.org
Date: 2017-04-28T06:13:29Z
Lists: pgsql-hackers

Attachments

Thanks for taking a look.

On 2017/04/28 5:24, Robert Haas wrote:
> On Wed, Apr 26, 2017 at 9:30 PM, Amit Langote
> <Langote_Amit_f8@lab.ntt.co.jp> wrote:
>>> Do we need to update the documentation?
>>
>> Yes, I think we should.  How about as in the attached?
> 
> Looks reasonable, but I was thinking you might also update the section
> which contrasts inheritance-based partitioning with declarative
> partitioning.

It seems to me that there is no difference in behavior between
inheritance-based and declarative partitioning as far as statement-level
triggers are concerned (at least currently).  In both the cases, we fire
statement-level triggers only for the table specified in the command.

Maybe, we will fix things someday so that statement-level triggers will be
fired for all the tables in a inheritance hierarchy when the root parent
is updated or deleted, but that's currently not true.  We may never
implement that behavior for declarative partitioned tables though, so
there will be a difference if and when we implement the former.

Am I missing something?

>> By the way, code changes I made in the attached are such that a subsequent
>> patch could implement firing statement-level triggers of all the tables in
>> a partition hierarchy, which it seems we don't want to do.  Should then
>> the code be changed to not create ResultRelInfos of all the tables but
>> only the root table (the one mentioned in the command)?  You will see that
>> the patch adds fields named es_nonleaf_result_relations and
>> es_num_nonleaf_result_relations, whereas just es_root_result_relation
>> would perhaps do, for example.
> 
> It seems better not to create any ResultRelInfos that we don't
> actually need, so +1 for such a revision to the patch.

OK, done.  It took a bit more work than I thought.

Updated patch attached.

Thanks,
Amit

Commits

  1. Fix typo.

  2. Document trigger-firing behavior for inheritance/partitioning.

  3. Fire per-statement triggers on partitioned tables.

  4. Set ecxt_scantuple correctly for tuple routing.

  5. Fix interaction of partitioned tables with BulkInsertState.

  6. Avoid core dump for empty prepared statement in an aborted transaction.

  7. Fix some problems in check_new_partition_bound().

  8. Remove unnecessary arguments from partitioning functions.

  9. Fix reporting of constraint violations for table partitioning.

  10. Fix tuple routing in cases where tuple descriptors don't match.

  11. Invalid parent's relcache after CREATE TABLE .. PARTITION OF.

  12. Doc: improve documentation about inheritance.