Simon Riggs <simon at 2ndquadrant.com> writes:
>> In principle, it ought to be possible.
>
> Good, thanks.
>
>> The one problem I'd foresee is that indices aren't shared across such
>> inheritance scenarios, so if the "child" tables get large, queries
>> across the "parent" might wind up not being as efficient as you'd
>> want.
>
> Are you saying a slony target table can't have an index on it?

No, certainly not.  You can have plenty of indices.

> Why would you not put an index on both pieces of the inherited table?

The problem is the "other" side of that, which is that queries that
aren't carefully constructed may not use the indices as nicely as
you'd hope.

We had "rotor table" queries (not involving inheritance) where
self-joins on a view turned into 81-way joins that were all Seq Scans
because there weren't any range restrictions on the pieces of the
view...  The resulting query ground to a halt :-(.  This wasn't a
Slony-I issue, per se...
-- 
"cbbrowne","@","ca.afilias.info"
<http://dev6.int.libertyrms.com/>
Christopher Browne
(416) 673-4124 (land)