[ getting back to this after assorted distractions ]
Andres Freund writes:
> On Thursday, January 12, 2012 02:24:44 AM Tom Lane wrote:
>> Andres Freund writes:
>>> On Thursday, January 12, 2012 01:01:01 AM Tom Lane wrote:
Looks pretty bogus to me. You're essentially assuming that the side
On Wed, 11 Jan 2012 19:40:34 -0500, Andres wrote:
> Unfortunately the pastebins where raptelan provided plans expired by
> now... Perhaps he can provide them again?
Sure, the original curiosity I noticed was that adjusting the block
size of results returned by the CTE had widely different effects,
On Thursday, January 12, 2012 02:24:44 AM Tom Lane wrote:
> Andres Freund writes:
> > On Thursday, January 12, 2012 01:01:01 AM Tom Lane wrote:
> >> Looks pretty bogus to me. You're essentially assuming that the side of
> >> the join without statistics is unique, which is a mighty dubious
> >> as
Andres Freund writes:
> On Thursday, January 12, 2012 01:01:01 AM Tom Lane wrote:
>> Looks pretty bogus to me. You're essentially assuming that the side of
>> the join without statistics is unique, which is a mighty dubious
>> assumption.
> It sure is a bit dubious. But assuming that a semijoi
On Thursday, January 12, 2012 01:01:01 AM Tom Lane wrote:
> (In cases where we know it's unique, something like this
> could be reasonable, but I believe get_variable_numdistinct already
> accounts for such cases.)
One of those case which looks relatively easy is that CTEs currently work as a
kin
On Thursday, January 12, 2012 01:01:01 AM Tom Lane wrote:
> Andres Freund writes:
> Also, why the asymmetry in null handling? And why did you only touch
> one of the two code paths in eqjoinsel_semi? They have both got this
> issue of how to estimate with inadequate stats.
This patch was purely
Andres Freund writes:
> Whats your opinion on this?
Looks pretty bogus to me. You're essentially assuming that the side of
the join without statistics is unique, which is a mighty dubious
assumption. (In cases where we *know* it's unique, something like this
could be reasonable, but I believe g
Hi all,
Monday RhodiumToad/Andrew Gierth and I tried to debug a plan of raptelan
(CCed) getting rather strange plans. After trawling through some unrelated
stuff we diagnosed that the problem were some rather strange estimates.
I managed to extract a reproducable, simple testcase. Our analysis