| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> |
|---|---|
| To: | David Rowley <david(dot)rowley(at)2ndquadrant(dot)com> |
| Cc: | Alvaro Herrera <alvherre(at)2ndquadrant(dot)com>, Robert Haas <robertmhaas(at)gmail(dot)com>, "pgsql-hackers(at)postgresql(dot)org" <pgsql-hackers(at)postgresql(dot)org> |
| Subject: | Re: Rethinking representation of partial-aggregate steps |
| Date: | 2016-06-27 15:00:28 |
| Message-ID: | 429.1467039628@sss.pgh.pa.us |
| Views: | Raw Message | Whole Thread | Download mbox | Resend email |
| Thread: | |
| Lists: | pgsql-hackers |
David Rowley <david(dot)rowley(at)2ndquadrant(dot)com> writes:
> I can't help wonder how plan to allow future expansions of
> non-serialized partial aggregates giving that in setrefs.c you're
> making a hard assumption that mark_partial_aggref() should always
> receive the SERIAL versions of the aggsplit.
What I was imagining, but didn't bother to implement immediately, is
that we could pass down the appropriate AggSplit value from the plan
node (using the context argument for the mutator function). planner.c
likewise needs a bit more plumbing to generate such plan nodes in the
first place, so I didn't feel that setrefs.c had to be smarter right now.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Tom Lane | 2016-06-27 15:03:39 | Re: Parallelized polymorphic aggs, and aggtype vs aggoutputtype |
| Previous Message | Tom Lane | 2016-06-27 14:13:47 | Broken handling of lwlocknames.h |