On Tue, Jul 17, 2012 at 6:08 PM, Simon Riggs
<simon@2ndquadrant.com> wrote:
On 17 July 2012 16:54, David E. Wheeler <
david@justatheory.com> wrote:
> Yeah, but that index is unnecessarily big if one will never use c or d in the search. The nice thing about covering indexes as described for SQLite 4 and implemented in MSSQL is that you can specify additional columns that just come along for the ride, but are not part of the indexed data:
>
> CREATE INDEX cover1 ON table1(a,b) COVERING(c,d);
>
> Yes, you can do that by also indexing c and d as of 9.2, but it might be nice to be able to include them in the index as additional row data without actually indexing them.
ISTM that it is a non-feature if the index is already non-unique, and
the difference is simply down to the amount of snake oil applied to
the descriptive text on the release notes.
It would be useful in non-unique indexes to store data without ordering operators (like xml).
