Justin Pryzby <pryzby@telsasoft.com> writes:
> On Fri, Jul 01, 2022 at 07:19:11PM -0400, Tom Lane wrote:
>> That, um, piqued my interest. After a bit of digging,
>> I modestly propose the attached. I'm not sure if it's
>> okay to back-patch this, because maybe someone out there
>> is relying on qsort() to be incapable of throwing an error
>> --- but it'd solve the problem at hand and a bunch of other
>> issues of the same ilk.
> Confirmed this fixes the 3 types of extended stats, at least for the example
> cases I made.
> If it's okay to backpatch, v14 seems adequate since the problem is more
> prominent with expressional statistics (I'm sure that's why I hit it) ...
> otherwise v15 would do.
After thinking for awhile, my inclination is to apply my patch in
HEAD and yours in v15/v14. We have a year to find out if there's
any problem with the more invasive check if we do it in HEAD;
but there's a lot less margin for error in v14, and not that
much in v15 either.
regards, tom lane