Re: Saner interval hash function

Greg Stark <stark(at)enterprisedb(dot)com> writes:
> On Fri, Apr 3, 2009 at 10:46 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>> The present implementation of interval_hash() is very carefully designed
>> and coded ... to meet the wrong specification :-(. What it should
>> be doing is producing equal hashcodes for values that interval_eq()
>> considers equal. The error is exhibited in the recent bug report #4748.

> It would be nice if we had a way to generate a lot of similar values
> for a every data type. Then we could have a regression test which
> checks for each data type that the hash function matches the equality
> operator -- and for that matter that the various inequality operators
> are also consistent.

> I'm not sure how to generate values though. For a lot of data types it
> would be hard to generate values densely enough to trigger any bugs.

Yeah. I did add a regression test for the specific case of '30 days'
vs '1 month', which we know is a pain point for this particular data
type. Generating values at random doesn't seem like it's really likely
to teach us much though.

regards, tom lane