On Tue, 2010-11-30 at 15:52 -0500, Robert Haas wrote:
> On Tue, Nov 30, 2010 at 2:45 PM, Jeff Davis <pgsql(at)j-davis(dot)com> wrote:
> > On Tue, 2010-11-30 at 11:12 -0500, Robert Haas wrote:
> >> Could you possibly have ModifyTable -> Limit -> MergeAppend?
> > Before MergeAppend knows which tuple to produce, it needs to see the
> > tuples (at least the first one from each of its children), meaning that
> > it needs to pull them through ModifyTable; and at that point it's
> > already too late.
> You seem to be imagining the MergeAppend node on top
Yes, I assumed that the tuples flowed in the direction of the arrows ;)
Now that I think about it, your representation makes some sense given
our EXPLAIN output.
In response to
pgsql-hackers by date
|Next:||From: Jeff Janes||Date: 2010-11-30 22:56:46|
|Subject: Re: Spread checkpoint sync|
|Previous:||From: Alvaro Herrera||Date: 2010-11-30 22:01:02|
|Subject: Re: DELETE with LIMIT (or my first hack)|