Peter Geoghegan <pg@heroku.com> writes:
> On Thu, Jun 23, 2016 at 4:08 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
>> So the answer is that this got broken by commit 9f03ca915196dfc8,
>> which appears to have imagined that _hash_form_tuple() is just an
>> alias for index_form_tuple(). But it ain't.
> Can we add test coverage that will detect when comparetup_index_hash()
> gives wrong answers, please? It's been broken at least 2 times, that
> I'm aware of. I'll write a patch, if that helps.
I agree that we need to do something, but I'm not very clear on what
the something is. I considered inventing a debug #define that would
reduce the _h_spool threshold to the minimum workable amount (2, looks
like), but I'm not sure how much that will help. What did you have
in mind?
regards, tom lane