On 06.10.2020 00:21, Pavel Borisov wrote:
> Hi, hackers!
> I added some extra tests for different cases of use of automatic
> partition creation.
> v3-0002 can be applied on top of the original v2 patch for correct
> work with some corner cases with constraints included in this test.
>
Thank you for the tests. I've added them and the fix into the patch.
I also noticed, that some table parameters, such as persistence were not
promoted to auto generated partitions. This is fixed now. The test cases
for temp and unlogged auto partitioned tables are updated respectively.
Besides, I slightly refactored the code and fixed documentation typos,
that were reported by Rahila.
With my recent changes, one test statement, that you've added as
failing, works.
CREATE TABLE list_parted_fail (a int) PARTITION BY LIST (a) CONFIGURATION
(VALUES IN ('1' collate "POSIX"));
It simply ignores collate POSIX part and creates a table with following
structure:
Partitioned table "public.list_parted_fail"
Column | Type | Collation | Nullable | Default | Storage | Stats
target | Description
--------+---------+-----------+----------+---------+---------+--------------+-------------
a | integer | | | | plain
| |
Partition key: LIST (a)
Partitions: list_parted_fail_0 FOR VALUES IN (1)
Do you think that it is a bug? For now, I removed this statement from
tests just to calm down the CI.
--
Anastasia Lubennikova
Postgres Professional: http://www.postgrespro.com
The Russian Postgres Company