David G Johnston wrote
> On Saturday, December 20, 2014, Mike Pultz [via PostgreSQL] <
> ml-node+s1045698n5831551h5@.nabble
>> wrote:
>
>> Hey David,
>>
>> > What version are you using?
>>
>> Sorry, I'm on 9.3.4.
>>
>> >Now() is a volatile function so the planner cannot omit partitions.
>> >
>> >Replace that with a constant and now it can.
>>
>> I'm not sure what you mean- now() is a stable function:
>
>
> My mistake but unless it is immutable the planner cannot evaluate it, it
> has to defer to the executor. The executor can evaluate it a single time
> for the query but it is still left with the original execution plan given
> to it by the planner.
>
>
>> it's the timestamp at the start of the transaction- so the planner should
>> have a set value for all rows.
>
>
>> Am I missing something else?
>>
>>
> Precise responsibility and timing mechanics between the planner and
> executor which I cannot give adequate exposition on off the top of my
> head...
>
> David J.
http://www.postgresql.org/docs/9.3/interactive/ddl-partitioning.html
Note the third-to-last paragraph.
David J.
--
View this message in context:
http://postgresql.nabble.com/Problem-with-constraint-exclusion-on-partitions-tp5831541p5831553.html
Sent from the PostgreSQL - general mailing list archive at Nabble.com.
