Re: Issue with point_ops and NaN

At Wed, 31 Mar 2021 15:46:16 +0900 (JST), Kyotaro Horiguchi <horikyota(dot)ntt(at)gmail(dot)com> wrote in
> At Wed, 31 Mar 2021 09:26:00 +0900, Michael Paquier <michael(at)paquier(dot)xyz> wrote in
> > On Tue, Mar 30, 2021 at 11:39:40PM +0800, Julien Rouhaud wrote:
> > > On Tue, Mar 30, 2021 at 11:02:32AM -0400, Tom Lane wrote:
> > >> Agreed --- one could make an argument for either 'false' or NULL
> > >> result, but surely not 'true'.
> > >
> > > I would think that it should return NULL since it's not inside nor outside the
> > > polygon, but I'm fine with false.
> >
> > Yeah, this is trying to make an undefined point fit into a box that
> > has a definition, so "false" does not make sense to me either here as
> > it implies that the point exists? NULL seems adapted here.
>
> Sounds reasonable. The function may return NULL for other cases so
> it's easily changed to NULL.
>
> # But it's bothersome to cover all parallels..

Hmm. Many internal functions handles bool, which cannot handle the
case of NaN naturally. In short, it's more invasive than expected.

> Does anyone oppose to make the case NULL? If no one objects, I'll do
> that.

Mmm. I'd like to reduce from +1 to +0.7 or so, considering the amount
of needed work...

regards.

--
Kyotaro Horiguchi
NTT Open Source Software Center