On Thu, Feb 09, 2012 at 12:02:29PM -0500, Robert Haas wrote:
When Heikki worked up his original index-only scan patches (which
didn't end up looking much like what eventually got committed), he had
the notion of an index-only qual. That is, given a query like this:
select sum(1) from foo
On Sat, Feb 11, 2012 at 7:16 AM, Jesper Krogh jes...@krogh.cc wrote:
Ok, but there are still cases where we don't even need to construct
a data tuple at all:
2012-02-11 13:14:01.579 jk=# explain select count(*) from testtable where
fts @@ to_tsquery('english','test1');
On 2012-02-09 22:17, Jesper Krogh wrote:
On 2012-02-09 21:09, Robert Haas wrote:
That doesn't make sense to me. If you probe index A for rows where a
= 1 and find that CTID (100,1) is such a row, and now want to return a
column value b that is not present in that index, the fastest way to
get
On 2012-02-09 18:02, Robert Haas wrote:
I don't have any appetite for trying to do anything more with
index-only scans for 9.2, though maybe someone else will think
otherwise. But I would like very much to get KaiGai's leakproof stuff
committed, and so it seems like a good idea to reconcile the
On Thu, Feb 9, 2012 at 1:33 PM, Jesper Krogh jes...@krogh.cc wrote:
On 2012-02-09 18:02, Robert Haas wrote:
I don't have any appetite for trying to do anything more with
index-only scans for 9.2, though maybe someone else will think
otherwise. But I would like very much to get KaiGai's
On 2012-02-09 21:09, Robert Haas wrote:
That doesn't make sense to me. If you probe index A for rows where a
= 1 and find that CTID (100,1) is such a row, and now want to return a
column value b that is not present in that index, the fastest way to
get the row is going to be to fetch block 100
