On 8 November 2018 at 20:15, Amit Langote <langote_amit...@lab.ntt.co.jp> wrote: > Actually, as I also proposed upthread, we should move root_tuple_slot from > PartitionTupleRouting to ModifyTableState as mt_root_tuple_slot, because > it's part of the first step described above that has nothing to do with > partition tuple routing proper. We can make PartitionTupleRouting private > to execPartition.c if we do that.
okay. Makes sense. I've changed things around to PartitionTupleRouting is now private to execPartition.c > I didn't find anything quite significant to complain about, except just > one line: > > + * Initially we must only setup 1 PartitionDispatch object; the one for > > setup -> set up Changed too. I've attached v15 and a delta from v14 to ease re-review. I also ran pgindent on this version. That's not part of the delta but is in the main patch. -- David Rowley http://www.2ndQuadrant.com/ PostgreSQL Development, 24x7 Support, Training & Services
v14_v15.diff
Description: Binary data
v15-0001-Speed-up-INSERT-and-UPDATE-on-partitioned-tables.patch
Description: Binary data
