On Mon, Oct 17, 2016 at 6:29 PM, Peter Geoghegan <pg@heroku.com> wrote:
> In light of the fact that the minimal fix has low overhead, I agree.
> I've drafted a patch to do this. I expect to be able to post it in the
> next few days.
I've put this off until I can get some buy-in from Heikki about a more
general issue with "should_free" arguments in tuplesort.c [1].
[1] https://www.postgresql.org/message-id/CAM3SWZQWZZ_N=DmmL7tKy_OUjGH_5mN=N=A6h7kHyyDvEhg2DA@mail.gmail.com
--
Peter Geoghegan