On Tue, Jan 10, 2017 at 4:17 AM, Amit Langote
<Langote_Amit_f8@lab.ntt.co.jp> wrote:
> On 2017/01/10 14:44, Keith Fiske wrote:
>> Is there any reason for the exclusion of parent tables from the pg_tables
>> system catalog view? They do not show up in information_schema.tables as
>> well. I believe I found where to make the changes and I tested to make sure
>> it works for my simple case. Attached is my first attempt at patching
>> anything in core. Not sure if there's anywhere else this would need to be
>> fixed.
>
> That's an oversight. The original partitioning patch didn't touch
> information_schema.sql and system_views.sql at all. I added the relkind =
> 'P' check in some other views as well, including what your patch considered.
I took a look at this patch:
* The SQL for pg_seclabels contained an obvious syntax error.
* pg_seclabels should only return object types that could be used in
the SECURITY LABEL command, so it needs to return 'table' not
'partitioned table' for the new relkind.
* There's a pointless change from v.relkind = 'v' to v.relkind IN ('v').
* I don't see any indication on the Internet that "PARTITIONED TABLE"
is a legal value for the table type in information_schema.tables.
Unless we can find something official, I suppose we should just
display BASE TABLE in that case as we do in other cases. I wonder if
the schema needs some broader revision; for example, are there
information_schema elements intended to show information about
partitions?
* Even though unique/primary key/foreign key constraints can't
currently be defined on partitioned tables, it seems best to change
the related reference symmetrically with the others, because we might
support it in the future and then this would break again. Similarly
for key_column_usage.
Committed with fixes for those issues.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company