Horton

Re: extended stats on partitioned tables

Tom Lane <tgl@sss.pgh.pa.us>

View thread View on postgr.es
From: Tom Lane <tgl@sss.pgh.pa.us>
To: Tomas Vondra <tomas.vondra@enterprisedb.com>
Cc: Zhihong Yu <zyu@yugabyte.com>, Justin Pryzby <pryzby@telsasoft.com>, Jaime Casanova <jcasanov@systemguards.com.ec>, PostgreSQL-development <pgsql-hackers@postgresql.org>
Date: 2021-12-12T18:27:22Z
Lists: pgsql-hackers 
Tomas Vondra <tomas.vondra@enterprisedb.com> writes:
> On 12/12/21 16:37, Zhihong Yu wrote:
>> Since the rte (RangeTblEntry*) doesn't seem to be used beyond checking 
>> inh, I think it would be better if the above style of checking is used 
>> throughout the patch (without introducing rte variable).

> It's mostly a matter of personal taste, but I always found this style of 
> condition (i.e. dereferencing a pointer returned by a function) much 
> less readable. It's hard to parse what exactly is happening, what struct 
> type are we dealing with, etc. YMMV but the separate variable makes it 
> much clearer for me. And I'd expect the compilers to produce pretty much 
> the same code too for those cases.

FWIW, I agree.  Also, it's possible that future patches would create a
need to touch the RTE again nearby, in which case having the variable
makes it easier to write non-crummy code for that.

			regards, tom lane

Commits

  1. Add stxdinherit flag to pg_statistic_ext_data

  2. Build inherited extended stats on partitioned tables

  3. Ignore extended statistics for inheritance trees

  4. Don't build extended statistics on inheritance trees

  5. Tighten up relation kind checks for extended statistics

  6. Avoid assuming that statistics for a parent relation reflect the properties of