Re: Planner hints in Postgresql

Atri Sharma wrote
> On Mon, Mar 17, 2014 at 9:45 PM, Tom Lane <

> tgl(at)(dot)pa

> > wrote:
>
>> David Johnston <

> polobo@

> > writes:
>> > Need to discuss the general "why" before any meaningful help on the
>> "how" is
>> > going to be considered by hackers.
>>
>> Possibly worth noting is that in past discussions, we've concluded that
>> the most sensible type of hint would not be "use this plan" at all, but
>> "here's what to assume about the selectivity of this WHERE clause".
>> That seems considerably less likely to break than any attempt to directly
>> specify plan details.
>>
>>
> Isnt using a user given value for selectivity a pretty risky situation as
> it can horribly screw up the plan selection?
>
> Why not allow the user to specify an alternate plan and have the planner
> assign a higher preference to it during plan evaluation? This shall allow
> us to still have a fair evaluation of all possible plans as we do right
> now
> and yet have a higher preference for the user given plan during
> evaluation?

The larger question to answer first is whether we want to implement
something that is deterministic...

How about just dropping the whole concept of "hinting" and provide a way for
someone to say "use this plan, or die trying." Maybe require it be used in
conjunction with named PREPAREd statements:

PREPARE s1 (USING /path/to/plan_def_on_server_or_something_similar) AS
SELECT ...;

Aside from whole-plan specification I can definitely see where join/where
specification could be useful if it can overcome the current limitation of
not being able to calculate inter-table estimations.

David J.

--
View this message in context: http://postgresql.1045698.n5.nabble.com/Planner-hints-in-Postgresql-tp5796347p5796378.html
Sent from the PostgreSQL - hackers mailing list archive at Nabble.com.