Re: [HACKERS] Proposal: Local indexes for partitioned table

On Tue, Jan 16, 2018 at 6:08 PM, David Rowley
<david.rowley@2ndquadrant.com> wrote:
> On 17 January 2018 at 03:58, Alvaro Herrera <alvherre@alvh.no-ip.org> wrote:
>>> 9. Error details claim that p2_a_idx is not a partition of p.
>>> Shouldn't it say table "p2" is not a partition of "p"?
>>
>> You missed the "on" in the DETAIL:
>>   DETAIL:  Index "p2_a_idx" is not on a partition of table "p".
>> You could argue that this is obscurely worded, but if you look at the
>> command:
>>    ALTER INDEX p_a_idx ATTACH PARTITION p2_a_idx;
>> nowhere is table p2 mentioned, so I'm not sure it's a great idea to
>> mention the table in the error message.
>
> I think I did miss the "on".

I think you will not be the only person to make that mistake.  I think
it would be better phrased as

DETAIL: "p2_a_idx" is not an index of any partition of table "p"

-- 
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company