Hi. On 2018/04/19 6:45, Alvaro Herrera wrote: > Amit Langote wrote: >> On Thu, Apr 19, 2018 at 12:01 AM, Alvaro Herrera >> <alvhe...@alvh.no-ip.org> wrote: > >>> Makes sense. Still, I was expecting that pruning of hash partitioning >>> would also work for pseudotypes, yet it doesn't. >> >> It does? > > Aha, so it does. > > While staring at this new code, I was confused as to why we didn't use > the commutator if the code above had determined one. I was unable to > cause a test to fail, so I put that thought aside.
Oops, you're right. Shouldn't have ignored the commutator. > Some time later, after restructuring the code in a way that seemed to > make more sense to me (and saving one get_op_opfamily_properties call > for the case of the not-equals operator), I realized that with the new > code we can store the opstrategy in the PartClause instead of leaving it > as Invalid and look it up again later, so I did that. And lo and > behold, the tests that used commutators started failing! So I fixed > that one in the obvious way, and the tests work fully again. > > Please give this version another look. I also rewrote a couple of > comments. Thanks, your rewritten version looks much better. > I now wonder if there's anything else that equivclass.c or indxpath.c > can teach us on this topic. I have referenced indxpath.c number of times when writing this code (for example, match_clause_to_indexcol), but never equivclass.c. Thanks, Amit
