On Tue, Mar 22, 2011 at 7:58 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> Throw error for indeterminate collation of an ORDER/GROUP/DISTINCT target.
>
hmm. In the current arrangement = doesn't depend on collation, right?
So in an ideal world we could use any collation we want to implement
GROUP/DISTINCT/UNION/INTERSECT/EXCEPT. I can see how it would be hard
to generate paths for every possible collation but it would be awfully
nice if we had a way to represent this so we could pick whichever
index we thought would be fastest instead of relying on the user to
select one.
--
greg