On Thu, Mar 1, 2018 at 6:57 AM, Amit Langote
<Langote_Amit_f8@lab.ntt.co.jp> wrote:
> On 2018/02/28 19:14, Ashutosh Bapat wrote:
>> On Wed, Feb 28, 2018 at 6:42 AM, Amit Langote wrote:
>>> BTW, should there be a relevant test in partition_join.sql? If yes,
>>> attached a patch (partitionwise-join-collation-test-1.patch) to add one.
>>
>> A partition-wise join path will be created but discarded because of
>> higher cost. This test won't see it in that case. So, please add some
>> data like other tests and add command to analyze the partitioned
>> tables. That kind of protects from something like that.
>
> Thanks for the review.
>
> Hmm, the added test is such that the partition collations won't match, so
> partition-wise join won't be considered at all due to differing
> PartitionSchemes, unless I'm missing something.
>
The point is we wouldn't know whether PWJ was not selected because of
PartitionScheme mismatch OR the PWJ paths were expensive compared to
non-PWJ as happens with empty tables. In both the cases we will see a
non-PWJ "plan" although in one case PWJ was not possible and in the
other it was possible. I think what we want to test is that PWJ Is not
possible with collation mismatch.
--
Best Wishes,
Ashutosh Bapat
EnterpriseDB Corporation
The Postgres Database Company