Re: Automatic optimization of IN clauses via INNER JOIN

On Thu, Dec 17, 2009 at 9:20 PM, Craig Ringer
<craig(at)postnewspapers(dot)com(dot)au> wrote:
> On 17/12/2009 11:57 PM, Tom Lane wrote:
>>
>> Thomas Hamilton<thomashamilton76(at)yahoo(dot)com>  writes:
>>>
>>> But in our testing under the same optimization and conditions INNER JOIN
>>> is significantly outperforming IN.
>>
>> [ shrug... ]  You haven't provided any details, so it's impossible to
>> offer any useful advice.
>
> In other words: can we discuss this with reference to a specific case?
> Please provide your queries, your EXPLAIN ANALYZE output, and other relevant
> details as per:
>
>  http://wiki.postgresql.org/wiki/SlowQueryQuestions
>
> I'd be interested in knowing whether the planner can perform such
> transformations and if so why it doesn't myself. I have the vague feeling
> there may be semantic differences in the handling of NULL but I can't
> currently seem to puzzle them out.

NOT IN is the only that really kills you as far as optimization is
concerned. IN can be transformed to a join. NOT IN forces a NOT
(subplan)-type plan, which bites - hard.

...Robert