Horton

Re: expanding inheritance in partition bound order

Amit Khandekar <amitdkhan.pg@gmail.com>

View thread View on postgr.es
From: Amit Khandekar <amitdkhan.pg@gmail.com>
To: Amit Langote <Langote_Amit_f8@lab.ntt.co.jp>
Cc: Robert Haas <robertmhaas@gmail.com>, Ashutosh Bapat <ashutosh.bapat@enterprisedb.com>, Pg Hackers <pgsql-hackers@postgresql.org>
Date: 2017-09-13T10:16:27Z
Lists: pgsql-hackers 
On 13 September 2017 at 15:32, Amit Langote
<Langote_Amit_f8@lab.ntt.co.jp> wrote:
> On 2017/09/11 18:56, Amit Langote wrote:
>> Attached updated patch does it that way for both partitioned table indexes
>> and leaf partition indexes.  Thanks for pointing it out.
>
> It seems to me we don't really need the first patch all that much.  That
> is, let's keep PartitionDispatchData the way it is for now, since we don't
> really have any need for it beside tuple-routing (EIBO as committed didn't
> need it for one).  So, let's forget about "decoupling
> RelationGetPartitionDispatchInfo() from the executor" thing for now and
> move on.
>
> So, attached is just the patch to make RelationGetPartitionDispatchInfo()
> traverse the partition tree in depth-first manner to be applied on HEAD.
>
> Thoughts?

+1. If at all we need the decoupling later for some reason, we can do
that incrementally.

Will review your latest patch by tomorrow.


-- 
Thanks,
-Amit Khandekar
EnterpriseDB Corporation
The Postgres Database Company

Commits

  1. Make RelationGetPartitionDispatchInfo expand depth-first.

  2. Expand partitioned tables in PartDesc order.

  3. Don't lock tables in RelationGetPartitionDispatchInfo.

  4. Speed up dropping tables with many partitions.