Re: BUG: pg_stat_statements query normalization issues with combined queries

From: Fabien COELHO <coelho(at)cri(dot)ensmp(dot)fr>
To: Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>
Cc: Kyotaro HORIGUCHI <horiguchi(dot)kyotaro(at)lab(dot)ntt(dot)co(dot)jp>, craig(dot)ringer(at)2ndquadrant(dot)com, robertmhaas(at)gmail(dot)com, PostgreSQL Developers <pgsql-hackers(at)postgresql(dot)org>
Subject: Re: BUG: pg_stat_statements query normalization issues with combined queries
Date: 2016-12-30 14:10:42
Message-ID: alpine.DEB.2.20.1612301453280.32017@lancre
Views: Raw Message | Whole Thread | Download mbox | Resend email
Lists: pgsql-hackers


>> Yeah, that's what I was thinking of. There aren't very many places that
>> would need to know about that, I believe; [...]
> For fixing the information in pg_stat_statement, the location data must be
> transported from the parsed node to the query to the planned node, because
> the later two nodes types are passed to different hooks.
> Now the detail is that utility statements, which seems to be nearly all of
> them but select/update/delete/insert, do not have plans: The statement itself
> is its own plan... so there is no place to store the location & length.

Here is an updated version:

Changes wrt v2:

- I have added the missing stuff under /nodes/, this is stupid code that
should be automatically generated:-(

- I have added comments in "gram.y" about how the length is computed.
I have also slightly simplified the rule code there.

- I have rename "location" in create table space to "location_dir"
to avoid confusion.

- I have renamed the fields "location" and "length" instead of q*.

- I have moved the location & lenth copies in standard_planner.

- I have fixed the function declaration typo.

- I have simplified pgss_store code to avoid a copy, and move the
length truncation in qtext_store.

- I have improved again the pg_stat_statement regression tests with
combined utility statement tests, which implied some fixes to
extract the right substring for utility queries.

However, not changed:

- I cannot use the intermediate node trick suggested by Tom because
it does not work for utility statements which do not have plans, so
the code still adds location & length, sorry.

- I still use the 'last_semicolon' lexer variable. The alternative is to
change rules so as not to skip empty statements, then write a loop to
compute the length based on successor location, and remove the empty
statements. It can be done, I do not think it is better, it is only
different and more verbose. I'll do it if required by a committer.


Attachment Content-Type Size
parsenodes-3.patch text/x-diff 126.3 KB

In response to


Browse pgsql-hackers by date

  From Date Subject
Next Message Yuriy Zhuravlev 2016-12-30 14:10:45 Re: WIP: About CMake v2
Previous Message Pavel Stehule 2016-12-30 14:07:56 Re: proposal: session server side variables