On Sun, Dec 6, 2020 at 9:09 AM Andy Fan <zhihui.fan1213@gmail.com> wrote:
>>
>> I have not been following this thread closely enough to understand
>> why we need a new "UniqueKeys" data structure at all.
>
>
> Currently the UniqueKey is defined as a List of Expr, rather than EquivalenceClasses.
> A complete discussion until now can be found at [1] (The messages I replied to also
> care a lot and the information is completed). This patch has stopped at this place for
> a while, I'm planning to try EquivalenceClasses, but any suggestion would be welcome.
>
>>
>> But if the
>> motivation is only to remove this overspecification, I humbly suggest
>> that it ain't worth the trouble.
AFAIK, the simple answer is we need some way to tell that certain
expressions together form a unique key for a given relation. E.g.
group by clause forms a unique key for the output of GROUP BY.
Pathkeys have a stronger requirement that the relation is ordered on
that expression, which may not be the case with uniqueness e.g. output
of GROUP BY produced by hash grouping. To me it's Pathkeys - ordering,
so we could use Pathkeys with reduced strength. But that might affect
a lot of places which depend upon stronger pathkeys.
--
Best Wishes,
Ashutosh Bapat