Ok, as always there's a lot more to take into account then when just
superficially looking at it.
And indeed your counterexample shows that you'd have to include all the
previous when-conditions too as false
WHERE x=0 IS DISTINCT FROM true AND 1/x > 100, which could become quite
messy (especially w
Thanks for your thoughts.
For the case where it isn't known if the case expression itself is indexed,
technically that should be added as a decision-node in the query planner.
After all there are 2 possibilities to handle that so it should be up to
the planner to choose the cheapest.
Having said
Paul van der Linden writes:
> For the case where it isn't known if the case expression itself is indexed,
> technically that should be added as a decision-node in the query planner.
That'd be fairly hard to do, if we're regarding this as an expression
simplification step, since expression simplif
David Rowley writes:
> On Fri, 5 Jun 2020 at 14:41, Paul van der Linden
> wrote:
>> If I have a query like:
>>
>> SELECT * FROM (
>> SELECT
>> CASE
>> WHEN field='value1' THEN 1
>> WHEN field='value2' THEN 2
>> END AS category
>> FROM table1
>> ) AS foo
>> WHERE category=1
>>
>> doesn't use the
On Fri, 5 Jun 2020 at 14:41, Paul van der Linden
wrote:
> If I have a query like:
>
> SELECT * FROM (
> SELECT
> CASE
> WHEN field='value1' THEN 1
> WHEN field='value2' THEN 2
> END AS category
> FROM table1
> ) AS foo
> WHERE category=1
>
> doesn't use the index on field, while te
