On Wed, Apr 30, 2014 at 11:03 AM, Jeff Janes <jeff.janes@gmail.com> wrote:
> If we don't put in the work to make them useful, then they won't ever become
> useful.
>
> If we do put in the effort (and it would be considerable) then I think they
> will be. But you may be correct that the effort required would perhaps be
> better used in making btree even more better. I don't think we can conclude
> that definitively without putting in the work to do the experiment.
My argument doesn't hinge on there being more important work to do.
Rather, I simply don't think that there is never going to be a
compelling reason to use hash indexes in production. Apart from the
obvious inflexibility, consider what it takes to make index creation
fast - insertion-style building of indexes is much slower. Consider
multi-key indexes.
Now, I'm not telling anyone what to work on, and if someone wants to
make hash indexes WAL-logged to plug that hole, don't let me stop you.
It probably makes sense as a project to learn more about Postgres
internals. However, it would be unfair to not speak up given my
misgivings around the practical utility of hash indexes.
--
Peter Geoghegan