| From: | Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> | 
|---|---|
| To: | Aris Tsois <atsois(at)dblab(dot)ece(dot)ntua(dot)gr> | 
| Cc: | pgsql-hackers(at)postgresql(dot)org | 
| Subject: | Re: How to avoid redundant Sort operations? (pgsql 7.1.2) | 
| Date: | 2002-01-22 16:31:12 | 
| Message-ID: | 15444.1011717072@sss.pgh.pa.us | 
| Views: | Whole Thread | Raw Message | Download mbox | Resend email | 
| Thread: | |
| Lists: | pgsql-hackers | 
Aris Tsois <atsois(at)dblab(dot)ece(dot)ntua(dot)gr> writes:
> Is there any way to avoid those redundant sort operations?
Information about sort order doesn't propagate up through "subquery
scan" nodes.  There are a couple of reasons for this, one being that
pathkeys aren't returned as part of a completed Plan; there'd have to be
some uglification of the APIs of subquery_planner, grouping_planner, etc
to return the appropriate pathkeys along with the finished plan.
A more subtle problem is that pathkeys of a sub-plan aren't necessarily
meaningful for an outer plan (eg, the variables being sorted on might
not even be visible to the outer query).  I think you'd have to do some
kind of translation to see if you could restate the inner pathkeys in
terms of the subplan's output variables.
regards, tom lane
| From | Date | Subject | |
|---|---|---|---|
| Next Message | Justin Clift | 2002-01-22 16:33:26 | Re: [HACKERS] PostgreSQL Licence: GNU/GPL | 
| Previous Message | Thomas Lockhart | 2002-01-22 16:25:11 | Re: [HACKERS] PostgreSQL Licence: GNU/GPL |