Re: dropping partitioned tables without CASCADE

From: Ashutosh Bapat <ashutosh(dot)bapat(at)enterprisedb(dot)com>
To: Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org>
Cc: Amit Langote <Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp>, Rushabh Lathia <rushabh(dot)lathia(at)gmail(dot)com>, Simon Riggs <simon(at)2ndquadrant(dot)com>, Pg Hackers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: dropping partitioned tables without CASCADE
Date: 2017-11-06 06:51:13
Message-ID: CAFjFpReLSRw8Pu5ELsRR2dosF4BTaRbcq2GzOhrp35uKE=5jEg@mail.gmail.com
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Somehow the earlier patches missed qualifying pg_get_expr() by
pg_catalog. Fixed it along with annotating the partitioned partition
as ", PARTITIONED".

On Fri, Nov 3, 2017 at 6:09 PM, Alvaro Herrera <alvherre(at)alvh(dot)no-ip(dot)org> wrote:
>>
>> Right now, we could do that if we order the list by bound expression;
>> lexically DEFAULT would come before FOR VALUES ... . But that's not
>> future-safe; we may have a bound expression starting with A, B or C.
>> Beyond that it really gets tricky to order the partitions by bounds.
>
> I was just thinking in changing the query to be "order by
> is_the_default_partition, partition_name" instead of just "order by
> partition_name". Sorting by bounds rather than name (a feature whose
> worth should definitely be discussed separately IMV) sounds a lot more
> complicated.

Right now we don't have a catalog column or a SQL function which can
tell whether a given partition is default partition based on the
partition bounds or otherwise. That's what it seemed when you
suggested ordering by "is_the_default_partition". Instead I have
ordered the partitions by pg_catalog.pg_get_expr(...) = 'DEFAULT'. We
can introduce a SQL function which takes child and parent oids and
return true if it's default partition and use that here, but that
seems more than what you are suggesting here. I have added that as a
separate patch.

If we tackle the problem of listing partitions by their bounds
somehow, DEFAULT partition listing would be tackled anyway. So, may be
we should leave it outside the scope of this patch.

--
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company

Attachment Content-Type Size
0001-Improve-d-output-of-a-partitioned-table.patch text/x-patch 8.1 KB
0002-Separate-default-partition-from-rest-of-the-partitio.patch text/x-patch 2.0 KB

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Noah Misch 2017-11-06 08:07:52 Race to build pg_isolation_regress in "make -j check-world"
Previous Message Masahiko Sawada 2017-11-06 06:31:06 Re: WIP: long transactions on hot standby feedback replica / proof of concept