Re: More stable query plans via more predictable column statistics

<p dir="ltr">On Apr 5, 2016 00:31, "Tom Lane" <<a href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>>
wrote:<br/> ><br /> > Alex Shulgin <<a href="mailto:alex.shulgin@gmail.com">alex.shulgin@gmail.com</a>>
writes:<br/> > > On Mon, Apr 4, 2016 at 1:06 AM, Tom Lane <<a
href="mailto:tgl@sss.pgh.pa.us">tgl@sss.pgh.pa.us</a>>wrote:<br /> > >> I'm inclined to<br /> > >>
revertthe aspect of 3d3bf62f3 that made us work from "d" (the observed<br /> > >> number of distinct values in
thesample) rather than stadistinct (the<br /> > >> extrapolated estimate for the table).  On reflection I
thinkthat that's<br /> > >> inconsistent with the theory behind the old MCV-cutoff rule.  It wouldn't<br />
>>> matter if we were going to replace the cutoff rule with something else,<br /> > >> but it's
beginningto sound like that won't happen for 9.6.<br /> ><br /> > > Please feel free to do what you think is
inthe best interest of the people<br /> > > preparing 9.6 for the freeze.  I'm not all that familiar with the
process,<br/> > > but I guess reverting this early might save some head-scratching if some<br /> > >
interestinginteractions of this change combined with some others are going<br /> > > to show up.<br /> ><br />
>I've reverted that bit; so we still have the improvements associated with<br /> > ignoring nulls, but nothing
elseat the moment.  I'll set this commitfest<br /> > item back to Waiting on Author, just in case you are able to
makesome<br /> > more progress before the end of the week.<p dir="ltr">OK, though it's unlikely that I'll get
productiveagain before next week, but maybe someone who has also been following this thread wants to step in?<p
dir="ltr">Thanks.<br/> --<br /> Alex<br />