Re: pg_dump emits ALTER TABLE ONLY partitioned_table

Amit,

* Amit Langote (Langote_Amit_f8(at)lab(dot)ntt(dot)co(dot)jp) wrote:
> OK, I agree. I tweaked the existing bullet point about differences from
> traditional inheritance when using ONLY with partitioned tables.

Please take a look at the attached and let me know your thoughts on it.
I changed the code to complain again regarding TRUNCATE ONLY, since that
never actually makes sense on a partitioned table, unlike ALTER TABLE
ONLY, and adjusted the error messages and added hints.

> Updated patches attached (0002 and 0003 unchanged).

Regarding these, 0002 changes pg_dump to handle partitions much more
like inheritance, which at least seems like a decent idea but makes me a
bit concerned about making sure we're doing everything correctly. In
particular, we should really have regression tests for non-inherited
CHECK (and NOT NULL) constraints on partitions. Perhaps there are tests
covering those cases in the standard regression suite, but it's not
obvious from the SQL.

Thanks!

Stephen