On Wed, Feb 16, 2022 at 2:29 AM John Naylor
<john.naylor@enterprisedb.com> wrote:
> Does anyone see a reason not to put in the necessary work to try it out?
Seems reasonable to me. It's always a bit difficult, I feel, to know
what test cases to use - almost any idea is going to have some case
where it's worse than what we do today, and there can always be some
user who does that exact thing 100% of the time. Moreover, it's hard
to be certain that test cases we construct - say, ordered data,
reverse ordered data, randomly ordered data, almost ordered data with
a single element out of place, etc. - are actually covering all of the
interesting cases. At the same time, I don't think anyone would
seriously disagree with what you say in the subject line, and we won't
make any progress by NOT trying things that are recommended in the
academic literature.
--
Robert Haas
EDB: http://www.enterprisedb.com