Re: unnesting multirange data types

On Sat, Jun 19, 2021 at 7:35 PM Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Alexander Korotkov <aekorotkov(at)gmail(dot)com> writes:
> > Patch successfully passed commitfest.cputube.org. I'm going to
> > re-push it if there are no objections.
>
> I'm still not happy about the way you've done the multirange-to-array
> part. I think we'd be better off improving the polymorphism rules so
> that that can be handled by one polymorphic function. Obviously that'd
> be a task for v15, but we've already concluded that just having the
> unnest ability would be minimally sufficient for v14.
>
> So I think you should trim it down to just the unnest part.

I'm not entirely sure it's worth introducing anyrangearray. There
might be not many use-cases of anyrangearray besides this cast
(normally one should use multirange instead of an array of ranges).
But I agree that this subject should be carefully considered for v15.

> In any case, beta2 wraps on Monday, and there is very little time
> left for a full round of buildfarm testing. I almost feel that
> it's too late to consider pushing this today. Tomorrow absolutely
> is too late for beta2.

+1
I also don't feel comfortable hurrying with unnest part to beta2.
According to the open items wiki page, there should be beta3. Does
unnest part have a chance for beta3?

------
Regards,
Alexander Korotkov