Re: [HACKERS] path toward faster partition pruning

On Sat, Apr 7, 2018 at 8:55 AM, David Rowley
<david.rowley@2ndquadrant.com> wrote:
> On 7 April 2018 at 15:14, Ashutosh Bapat
> <ashutosh.bapat@enterprisedb.com> wrote:
>> On Sat, Apr 7, 2018 at 8:37 AM, David Rowley
>>> Why is writing tests that produce the same output required?
>>>
>>> We have many tests with alternative outputs. Look in
>>> src/tests/regress/expected for files matching _1.out
>>>
>>
>> That's true, but we usually add such alternative output when we know
>> all the variants possible as long as "all the variants" do not cover
>> everything possible. AFAIU, that's not true here. Also, on a given
>> machine a particular row is guaranteed to fall in a given partition.
>> On a different machine it will fall in some other partition, but
>> always that partition on that machine. We don't have a way to select
>> alternate output based on the architecture. May be a better idea is to
>> use .source file, creating .out on the fly based on the architecture
>> of machine like testing the hash output for a given value to decide
>> which partition it will fall into and then crafting .out with that
>> partition's name.
>
> Sounds like you're saying that if we have too many alternative files
> then there's a chance that one could pass by luck.

Yes.

>
> Maybe we can just back up what's just been committed by actually
> executing the queries and ensuring that all rows that made it into the
> table make it back out again.

That's one way. But how would we make sure that they landed in proper
partition. Actually we do not know what's proper partition for a given
architecture. And how would we make sure that all rows with the same
partition key land in the same partition. That's why I am suggesting
to calculate the hash value, take modulo and craft the name of
partition where corresponding row will land on a given architecture.
That way, we are sure that the tuple routing logic is correct and also
the partition pruning logic.

-- 
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company