Horton

Re: PoC plpgsql - possibility to force custom or generic plan

Merlin Moncure <mmoncure@gmail.com>

View thread View on postgr.es
From: Merlin Moncure <mmoncure@gmail.com>
To: Robert Haas <robertmhaas@gmail.com>
Cc: Pavel Stehule <pavel.stehule@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, Peter Eisentraut <peter.eisentraut@2ndquadrant.com>, Simon Riggs <simon@2ndquadrant.com>, Daniel Gustafsson <daniel@yesql.se>, Andrew Dunstan <andrew.dunstan@2ndquadrant.com>, Andres Freund <andres@anarazel.de>, Petr Jelinek <petr.jelinek@2ndquadrant.com>, David Steele <david@pgmasters.net>, PostgreSQL Hackers <pgsql-hackers@postgresql.org>, Jim Nasby <Jim.Nasby@bluetreble.com>
Date: 2017-09-19T18:49:56Z
Lists: pgsql-hackers 
On Tue, Sep 19, 2017 at 1:37 PM, Robert Haas <robertmhaas@gmail.com> wrote:
> On Tue, Sep 19, 2017 at 12:45 PM, Pavel Stehule <pavel.stehule@gmail.com> wrote:
>>> You can already set a GUC with function scope.  I'm not getting your
>>> point.
>>
>> yes, it is true. But implementation of #option is limited to PLpgSQL - so
>> there is not any too much questions - GUC is global - there is lot of
>> points:
>>
>> * what is correct impact on PREPARE
>> * what is correct impact on EXECUTE
>> * what should be done if this GUC is changed ..
>
> For better or for worse, as a project we've settled on GUCs as a way
> to control behavior.  I think it makes more sense to try to apply that
> option to new behaviors we want to control than to invent some new
> system.

This seems very sensible.

We also have infrastructure at the SQL level (SET) to manage the GUC.
Tom upthread (for pretty good reasons) extending SET to pl/pgsql
specific scoping but TBH I'm struggling as to why we need to implement
new syntax for this; the only thing missing is being able to scope SET
statements to a code block FWICT.

merlin

Commits

  1. Add plan_cache_mode setting

  2. Add some noreturn attributes to help static analyzers

  3. document when PREPARE uses generic plans

  4. Redesign the plancache mechanism for more flexibility and efficiency.