Movatterモバイル変換

[0]ホーム
URL:

Skip to content

Navigation Menu

Sign in
Appearance settings

Search code, repositories, users, issues, pull requests...

Provide feedback

We read every piece of feedback, and take your input very seriously.

Saved searches

Use saved searches to filter your results more quickly

 Sign up
Appearance settings

Commit443df6e

Browse files
committed
Revert "Optimize order of GROUP BY keys".
This reverts commitdb0d67d andseveral follow-on fixes. The idea of making a cost-based choiceof the order of the sorting columns is not fundamentally unsound,but it requires cost information and data statistics that we don'treally have. For example, relying on procost to distinguish therelative costs of different sort comparators is pretty pointlessso long as most such comparator functions are labeled with cost 1.0.Moreover, estimating the number of comparisons done by Quicksortrequires more than just an estimate of the number of distinct valuesin the input: you also need some idea of the sizes of the largergroups, if you want an estimate that's good to better than a factor ofthree or so. That's data that's often unknown or not very reliable.Worse, to arrive at estimates of the number of calls made to thelower-order-column comparison functions, the code needs to makeestimates of the numbers of distinct values of multiple columns,which are necessarily even less trustworthy than per-column stats.Even if all the inputs are perfectly reliable, the cost algorithmas-implemented cannot offer useful information about how to ordersorting columns beyond the point at which the average group sizeis estimated to drop to 1.Close inspection of the code added bydb0d67d shows that thereare also multiple small bugs. These could have been fixed, butthere's not much point if we don't trust the estimates to beaccurate in-principle.Finally, the changes in cost_sort's behavior made for very largechanges (often a factor of 2 or so) in the cost estimates for allsorting operations, not only those for multi-column GROUP BY.That naturally changes plan choices in many situations, and there'sprecious little evidence to show that the changes are for the better.Given the above doubts about whether the new estimates are reallytrustworthy, it's hard to summon much confidence that these changesare better on the average.Since we're hard up against the release deadline for v15, let'srevert these changes for now. We can always try again later.Note: in v15, I left T_PathKeyInfo in place in nodes.h even thoughit's unreferenced. Removing it would be an ABI break, and it seemsa bit late in the release cycle for that.Discussion:https://postgr.es/m/TYAPR01MB586665EB5FB2C3807E893941F5579@TYAPR01MB5866.jpnprd01.prod.outlook.com
1 parentb507a7a commit443df6e

File tree

24 files changed

+453
-1867
lines changed

24 files changed

+453
-1867
lines changed

‎contrib/postgres_fdw/expected/postgres_fdw.out

Lines changed: 9 additions & 6 deletions
Original file line numberDiff line numberDiff line change
@@ -2840,13 +2840,16 @@ select c2 * (random() <= 1)::int as c2 from ft2 group by c2 * (random() <= 1)::i
28402840
-- GROUP BY clause in various forms, cardinal, alias and constant expression
28412841
explain (verbose, costs off)
28422842
select count(c2) w, c2 x, 5 y, 7.0 z from ft1 group by 2, y, 9.0::int order by 2;
2843-
QUERY PLAN
2844-
------------------------------------------------------------------------------------------------------------
2845-
Foreign Scan
2843+
QUERY PLAN
2844+
---------------------------------------------------------------------------------------
2845+
Sort
28462846
Output: (count(c2)), c2, 5, 7.0, 9
2847-
Relations: Aggregate on (public.ft1)
2848-
Remote SQL: SELECT count(c2), c2, 5, 7.0, 9 FROM "S 1"."T 1" GROUP BY 2, 3, 5 ORDER BY c2 ASC NULLS LAST
2849-
(4 rows)
2847+
Sort Key: ft1.c2
2848+
-> Foreign Scan
2849+
Output: (count(c2)), c2, 5, 7.0, 9
2850+
Relations: Aggregate on (public.ft1)
2851+
Remote SQL: SELECT count(c2), c2, 5, 7.0, 9 FROM "S 1"."T 1" GROUP BY 2, 3, 5
2852+
(7 rows)
28502853

28512854
select count(c2) w, c2 x, 5 y, 7.0 z from ft1 group by 2, y, 9.0::int order by 2;
28522855
w | x | y | z

‎doc/src/sgml/config.sgml

Lines changed: 0 additions & 14 deletions
Original file line numberDiff line numberDiff line change
@@ -5055,20 +5055,6 @@ ANY <replaceable class="parameter">num_sync</replaceable> ( <replaceable class="
50555055
</listitem>
50565056
</varlistentry>
50575057

5058-
<varlistentry id="guc-enable-groupby-reordering" xreflabel="enable_group_by_reordering">
5059-
<term><varname>enable_group_by_reordering</varname> (<type>boolean</type>)
5060-
<indexterm>
5061-
<primary><varname>enable_group_by_reordering</varname> configuration parameter</primary>
5062-
</indexterm>
5063-
</term>
5064-
<listitem>
5065-
<para>
5066-
Enables or disables reordering of keys in a <literal>GROUP BY</literal>
5067-
clause. The default is <literal>on</literal>.
5068-
</para>
5069-
</listitem>
5070-
</varlistentry>
5071-
50725058
<varlistentry id="guc-enable-hashagg" xreflabel="enable_hashagg">
50735059
<term><varname>enable_hashagg</varname> (<type>boolean</type>)
50745060
<indexterm>

0 commit comments

Comments
 (0)
[8]ページ先頭
©2009-2025 Movatter.jp