On Fri, Dec 8, 2017 at 7:36 AM, Robert Haas <robertmh...@gmail.com> wrote: > On Wed, Dec 6, 2017 at 1:05 AM, Amit Kapila <amit.kapil...@gmail.com> wrote: >> Right and seeing that I have prepared the patch (posted above [1]) >> which fixes it such that it will resemble the non-parallel case. > > Long story short, I like the patch.
LGTM. There might be an argument for clearing the instrumentation every time on the basis that you might finish up keeping data from a non-final loop when a worker opted not to do anything in the final loop, but I'm not going to make that argument because I don't think it matters. The patch makes the tests in test-hash-join-rescan-instr-v1.patch pass (from my previous message). Please also consider that test patch for commit. -- Thomas Munro http://www.enterprisedb.com
