Re: A rather hackish POC for alternative implementation of WITH TIES - Mailing list pgsql-hackers

From Andrew Gierth
Subject Re: A rather hackish POC for alternative implementation of WITH TIES
Date
Msg-id 878slznkf0.fsf@news-spur.riddles.org.uk
Whole thread Raw
In response to Re: A rather hackish POC for alternative implementation of WITH TIES  (Alvaro Herrera <alvherre@2ndquadrant.com>)
Responses Re: A rather hackish POC for alternative implementation of WITH TIES
Re: A rather hackish POC for alternative implementation of WITH TIES
List pgsql-hackers
>>>>> "Alvaro" == Alvaro Herrera <alvherre@2ndquadrant.com> writes:

 Alvaro> My own inclination is that Andrew's implementation, being more
 Alvaro> general in nature, would be the better one to have in the
 Alvaro> codebase; but we don't have a complete patch yet. Can we reach
 Alvaro> some compromise such as if Andrew's patch is not completed then
 Alvaro> we push Surafel's?

Mine needs some attention to where exactly in planning the necessary
transformation work should be done; right now the planner part is a
hack, intended to demonstrate the idea (and to let the executor changes
work) rather than actually be the final version. As I mentioned before,
some stuff does need to be pushed out to an InitPlan to make it work
without multiple-evaluation problems.

(A second opinion from another planner expert would be welcome on that
part)

I was largely holding off on doing further work hoping for some
discussion of which way we should go. If you think my approach is worth
pursuing (I haven't seriously tested the performance, but I'd expect it
to be slower than Surafel's - the price you pay for flexibility) then I
can look at it further, but figuring out the planner stuff will take
some time.

-- 
Andrew.



pgsql-hackers by date:

Previous
From: Justin Pryzby
Date:
Subject: Re: error context for vacuum to include block number
Next
From: Peter Geoghegan
Date:
Subject: Re: Index Skip Scan