> The individual runs have quite a bit of variation.
It's a known behavior of pgbench.
> It's good to cross check results against known facts. If my napkin
> math is right, a ~1% speedup for ~60ktps/s amounts to saving ~160ns
> per query. It's not plausible that forcing the compiler's hand for
>
On Mon, Sep 1, 2025 at 8:13 AM Tatsuo Ishii wrote:
>
> > Branch alignment surely shouldn't matter in a function that is called
> > once per query?
>
> According to my test, it seems using unlikely() makes a small but
> non-negligible performance improvement over the code without
> unlikely() for s
Hi Chao Li,
>> v5 patch looks good to me. If there's no objection, I plan to push the
>> patch in early September.
>>
>
> Hi Tatsuo san,
>
> Thank you much very for your support.
You are welcome.
Can you please rebase v5 patch? It does not apply to current master
anymore.
$ git apply ~/v5-00
> + if (unlikely(Debug_print_raw_parse))
>
> Branch alignment surely shouldn't matter in a function that is called
> once per query?
According to my test, it seems using unlikely() makes a small but
non-negligible performance improvement over the code without
unlikely() for small queries.
Test m
Hi Tatsuo san,
Rebased v6 is attached.
Chao Li (Evan)
-
HighGo Software Co., Ltd.
https://www.highgo.com/
Tatsuo Ishii 于2025年9月1日周一 10:13写道：
> Hi Chao Li,
>
> >> v5 patch looks good to me. If there's no objection, I plan to push the
> >> patch in early September.
> >>
> >
On Wed, Aug 20, 2025 at 8:26 AM Tatsuo Ishii wrote:
> Thanks for updating the patch.
> v5 patch looks good to me. If there's no objection, I plan to push the
> patch in early September.
+ if (unlikely(Debug_print_raw_parse))
Branch alignment surely shouldn't matter in a function that is called
o
> On Aug 20, 2025, at 09:25, Tatsuo Ishii wrote:
>>
> v5 patch looks good to me. If there's no objection, I plan to push the
> patch in early September.
>
Hi Tatsuo san,
Thank you much very for your support.
Best regards,
--
Chao Li (Evan)
HighGo Software Co., Ltd.
https://www.highgo.com/
> On 2025/8/18 17:13, Tatsuo Ishii wrote:
>> + if (unlikely(Debug_print_parse))
>>
>> This should be:
>>
>> +if (unlikely(Debug_print_raw_parse))
>
> My bad! Such a stupid mistake. Fixed. And I did a test by turning
> on/off the flag in postgres.conf, and raw parse tree is only dumped
> whe
On 2025/8/18 17:13, Tatsuo Ishii wrote:
+ if (unlikely(Debug_print_parse))
This should be:
+ if (unlikely(Debug_print_raw_parse))
My bad! Such a stupid mistake. Fixed. And I did a test by turning on/off
the flag in postgres.conf, and raw parse tree is only dumped when
"debug_print
>> --- a/src/backend/tcop/postgres.c
>> +++ b/src/backend/tcop/postgres.c
>> @@ -649,6 +649,10 @@ pg_parse_query(const char *query_string)
>> TRACE_POSTGRESQL_QUERY_PARSE_DONE(query_string);
>> + if (Debug_print_raw_parse)
>> + elog_node_display(LOG, "raw parse tree", raw_parsetree_list,
>>
Hi Tatsuo san,
Thank you very much for your review.
On 2025/8/16 13:56, Tatsuo Ishii wrote:
I have looked into this patch.
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 20ccb2d6b54..4370e8307f2 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
Aroun
> Updated the patch to v3 version.
I have looked into this patch.
diff --git a/doc/src/sgml/config.sgml b/doc/src/sgml/config.sgml
index 20ccb2d6b54..4370e8307f2 100644
--- a/doc/src/sgml/config.sgml
+++ b/doc/src/sgml/config.sgml
Around line 7407 of config.sgml:
These parameters enable
Updated the patch to v3 version.
Chao Li (Evan)
--
HighGo Software Inc.
https://www.highgo.com/
Chao Li 于2025年8月5日周二 14:25写道：
> CommitFests patch created: https://commitfest.postgresql.org/patch/5946/
>
> 2025年8月4日 16:17，Chao Li 写道：
>
> I just noticed that my IDE a
CommitFests patch created: https://commitfest.postgresql.org/patch/5946/
> 2025年8月4日 16:17，Chao Li 写道：
>
> I just noticed that my IDE auto formatted guc_tables.c, which generated a lot
> of unnecessary diffs. Recreated the patch, and attach the v2 version.
>
> Chao Li (Evan)
>
I just noticed that my IDE auto formatted guc_tables.c, which generated a
lot of unnecessary diffs. Recreated the patch, and attach the v2 version.
Chao Li (Evan)
--
HighGo Software Inc.
https://www.highgo.com/
Chao Li 于2025年8月4日周一 15:57写道：
> Oh, I forget to attache
Oh, I forget to attached the patch file, here comes it.
Chao Li (Evan)
--
HighGo Software Inc.
https://www.highgo.com/
Chao Li 于2025年8月4日周一 15:53写道：
> I was reviewing some patches today, and during debugging the patches, I
> wanted to view raw parse tree, so I had t
I was reviewing some patches today, and during debugging the patches, I
wanted to view raw parse tree, so I had to apply my local patch of dumping
raw parse for every review, which was so inconvenient.
You may argue that raw parse tree might not be useful for every reviews. I
am still ramping up P
Hi Tatsuo, thanks for pointing out the past conversation.
So, the requirement got 1 more vote from me. But to not make noise to
people who are not interested in raw parse tree, I guess it's better to add
a new option "debug_print_raw_parse". For people who are interested in raw
parse tree, turning
> Before submitting this trivial patch, I still want to confirm with the
> community if it's intentional to not dump raw parse tree?
For your reference, here's the past discussion:
https://www.postgresql.org/message-id/flat/20080730.172949.132921436.t-ishii%40sraoss.co.jp
Best regards,
--
Tatsuo
Chao Li 于2025年8月1日周五 13:44写道：
> Hi,
>
> When "debug_print_parse" is "on", only the Query structure tree is
> dumped, the raw parse tree is not dumped to log. In some cases, viewing raw
> parse trees are also helpful. It is very hard to view a tree by watching
> variables in a debugger, so I added
20 matches
Mail list logo
