Re: Making clausesel.c Smarter

On Mon, Apr 3, 2017 at 6:22 PM, David Rowley
<david(dot)rowley(at)2ndquadrant(dot)com> wrote:
> On 4 April 2017 at 08:24, Andres Freund <andres(at)anarazel(dot)de> wrote:
>> On 2017-04-03 20:59:42 +1200, David Rowley wrote:
>>> Updated patch attached.
>>>
>>> Thanks for reviewing it.
>>
>> Given the time in the release cycle I'm afraid that this it's too late
>> to get this into v10. Does anybody disagree? If not, it should be
>> moved to the next CF.
>
> I strongly disagree. The time in the release cycle is the final
> commitfest. This is when patches are commited to the repository. The
> exception to this is that no large invasive patches should arrive new
> in the final commitfest. This is not one of those.

Selectivity misestimations can have a huge impact on query
performance, and that could be a big deal if there's a big regression
on some queries.

But, the beta period could be enough to detect any issues there. I'll
leave that decision to committers or commitfest managers I guess.

>
> Tom has already mentioned he'd like to look at this. If you're not
> willing, then please just don't look at it, but please also don't
> remove other peoples opportunity for doing so.
>
> Please explain your logic for thinking otherwise. Have I somehow
> misunderstood what the 1 week extension means?

But Tom hasn't yet, and even if he does, we'd have to be done with the
review within this week.

In any case, I intend to go on with the review later today, and at
first look the patch seems in good shape (though it'll take a deeper
look before I make that my official review). Most other patches in
need of review are a bit complex for my knowledge of the code base,
and I already started reviewing this one. We could perhaps be done by
the end of the week.

We could defer that decision to within 4 days. If we're done, we're
done. If we're not, move to next CF.