Hi,
On 2018-03-27 13:14:15 +1300, David Rowley wrote:
> I have to say, it really would be a shame to have this concern block
> us from future optimisations in aggregation.
Yea, I think that's an important point. By the dint of Tom's argument
we're never going to be able to provide parallelism for any aggregate
where input order matters. It's not like users are magically going to
stop depending on it, if they already do. Seems more likely that we'll
grow more users depending on it.
We could theoretically provide differently named aggregates where one
allows parallelism, the other doesn't, without duplicating the backing
code... Not pretty tho.
Greetings,
Andres Freund