Hi,
On Fri, Sep 01, 2022 at 09:55:15PM +0200, Daniel Gustafsson wrote:
> > On 2 Sep 2022, at 21:22, Ranier Vilela <ranier.vf@gmail.com> wrote:
> >
> > Em sex., 2 de set. de 2022 às 16:15, Daniel Gustafsson <daniel@yesql.se <mailto:daniel@yesql.se>> escreveu:
> > > On 2 Sep 2022, at 21:08, Ranier Vilela <ranier.vf@gmail.com <mailto:ranier.vf@gmail.com>> wrote:
> >
> > > At function circle_same the second isnan test is wrong.
> >
> > Yeah, that seems pretty wrong. Did you attempt to procure a test for when this
> > yields the wrong result?
> > Hi Daniel,
> > Unfortunately not.
>
> On HEAD, the below query yields what seems to be the wrong result for the "same
> as" operator:
>
> postgres=# select '<(0,0),NaN>'::circle ~= '<(0,0),1>'::circle;
> ?column?
> ----------
> t
> (1 row)
>
> With the patch applied, it returns the expected:
>
> postgres=# select '<(0,0),NaN>'::circle ~= '<(0,0),1>'::circle;
> ?column?
> ----------
> f
> (1 row)
>
> There seems to be surprisingly few tests around these geo operators?
Yeah, there are unfortunately a lot of problems around those and NaN, with
multiple reports in the past (I recall [1] and [2] but there were others).
There was a CF entry that tried to improve things [3], part of it was committed
but not all [4], and clearly some more work is needed.
[1] https://www.postgresql.org/message-id/flat/CAGf+fX70rWFOk5cd00uMfa__0yP+vtQg5ck7c2Onb-Yczp0URA@mail.gmail.com
[2] https://www.postgresql.org/message-id/20210330095751.x5hnqbqcxilzwjlm@nol
[3] https://commitfest.postgresql.org/38/2710/
[4] https://www.postgresql.org/message-id/3558828.1659045056@sss.pgh.pa.us
