On Thu, Jul 18, 2019 at 11:18 AM Amit Langote <amitlangot...@gmail.com> wrote: > On Tue, Jul 16, 2019 at 8:22 PM Etsuro Fujita <etsuro.fuj...@gmail.com> wrote: > > On Tue, Jul 2, 2019 at 6:29 PM Amit Langote <amitlangot...@gmail.com> wrote: > > > 0001 - fix partitionwise join to work correctly with n-way joins of > > > which some are full joins (+ cosmetic improvements around the code > > > that was touched) > > > > Here are my comments about the cosmetic improvements: they seem pretty > > large to me, so I'd make a separate patch for that. > > OK, my bad that I added so many cosmetic changes into a patch that is > meant to fix the main issue. Just to clarify, I'm proposing these > cosmetic improvements to better clarify the terminological separation > between nullable and non-nullable partition keys, which I found a bit > hard to understand as is.
OK, thanks for the explanation! > I've broken the patch into two: 0001 contains only cosmetic changes > and 0002 the fix for handling full joins properly. Would you rather > that be reversed? I like this order. > > In addition, I'd > > move have_partkey_equi_join() and match_expr_to_partition_keys() to > > relnode.c, because these functions are only used in that file. > > I hadn't noticed that. Makes sense to move them to relnode.c, which > is implemented in 0001. Thanks for including that! Will review. Best regards, Etsuro Fujita