Re: pgsql: Make heap TID a tiebreaker nbtree index column.

Peter Geoghegan <pg(at)bowt(dot)ie> writes:
> On Sun, Mar 24, 2019 at 11:02 AM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> I think I can probably get that done today, but if you don't want to
>> wait, feel free to put in the detail-suppression for now.

> I'll monitor the situation, and proceed with a stopgap
> detail-suppression fix this evening if, for whatever reason, it seems
> necessary.

It turns out we can't use the ObjectAddresses infrastructure because
it doesn't store enough information. So that means an extra qsort
comparator function, which is slightly annoying, but it's still not
a huge amount of code.

This flips one expected result to another order from what it was.
(I experimented with OID-descending sort, but that flips two
other expected results.)

If no objections, I'll push shortly.

regards, tom lane