On 10 February 2017 at 08:18, Amit Langote
<Langote_Amit_f8@lab.ntt.co.jp> wrote:
> I agree that getting the proposed documentation changes committed would be
> a step ahead.
Committed. I tested how it works and added documentation that matched
my experiences, correcting what you'd said and adding more information
for clarity as I went.
Few points from tests
* "ERROR: cannot create index on partitioned table "measurement_year_month""
is misleading because you can create indexes on partitions
* You can create additional CHECK (column is NOT NULL) as a check
constraint, even if you can't create a not null constraint
* The OID inheritance needs work - you shouldn't need to specify a
partition needs OIDS if the parent has it already.
* There's no tab completion, which prevents people from testing this
(maybe better now with some docs)
* ERROR: no partition of relation "measurement_year_month" found for row
DETAIL: Failing row contains (2016-12-02, 1, 1).
should not return the whole row, just the partition keys
* It's very weird you can't DROP a partitioned table. I think you need
to add dependencies.
* ERROR: TO must specify exactly one value per partitioning column
should say something more like "you specified one column value,
whereas the partitioning key requires two columns"
Good work so far, but there is still a very significant amount of work
to do. And as this feature evolves it must now contain full
documentation at every step, otherwise it won't be able to receive
full and fair review. So please make sure each new patch contains docs
changes for that patch.
--
Simon Riggs http://www.2ndQuadrant.com/
PostgreSQL Development, 24x7 Support, Remote DBA, Training & Services
