Re: POC: GROUP BY optimization

On 06/07/2018 03:41 PM, Teodor Sigaev wrote:
>>
>> ... snip ...
>>
>>> Priorization of the user-provided order can be as simple as giving
>>> that comparison_cost a small handicap.
>>
>> I see no point in doing that, and I don't recall a single place in the
>> planner where we do that. If the user specified ORDER BY, we'll slap an
>> explicit Sort on top when needed (which acts as the handicap, but in a
>> clear way). Otherwise we don't do such things - it'd be just plain
>> confusing (consider "ORDER BY a,b" vs. "ORDER BY b,c" with same data
>> types, ndistinct etc. but unexpectedly different costs). Also, what
>> would be a good value for the handicap?
>
> Again agree. If we have fixed order of columns (ORDER BY) then we should
> not try to reorder it. Current patch follows that if I didn't a mistake.
>

This part seems to be more a misunderstanding between me and Claudio. I
believe Claudio was referring to the column order in a GROUP BY, not
ORDER BY. In which case we don't add any Sort, of course.

I'm still opposed to adding arbitrary handicap to prioritize the order
specified by user, for the reasons I explained before. We should aim to
make the heuristics/costing reliable enough to make this unnecessary.

regards

--
Tomas Vondra http://www.2ndQuadrant.com
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services