spoly that was valid now detected as invalid

[pdowler]

I have a polygon from an existing server that later versions of pgsphere detect as invalid.

Summary: The polygon below is acceptable in pgsphere up to 1.2.3 and then rejected by 1.3.1 and 1.4.1.

psql (15.4, server 15.5)
Type "help" for help.

content=> select pg_sphere_version();
 pg_sphere_version 
-------------------
 1.2.3
(1 row)

content=> select '{
(3.30474723646012 , 1.08600456205300),
(3.30341855309927 , 1.08577960186707),
(3.30341054542378 , 1.08578643990271),
(3.30297351563319 , 1.08633534556428),
(3.30357156120003 , 1.08643683957210),
(3.30358891855857 , 1.08643995044436),
(3.30360894676365 , 1.08644306147078),
(3.30361829343581 , 1.08644430596871),
(3.30362630482521 , 1.08644555030213),
(3.30364633346451 , 1.08644866102000),
(3.30365300940335 , 1.08645052692055),
(3.30366102096957 , 1.08645177113937),
(3.30367036769496 , 1.08645363721023),
(3.30367837934959 , 1.08645488137174),
(3.30368906174976 , 1.08645612569695),
(3.30370107936906 , 1.08645799183673),
(3.30370642025712 , 1.08645985750225),
(3.30373179124734 , 1.08646358962156),
(3.30374514456618 , 1.08646545561358),
(3.30410706158729 , 1.08652886672786),
(3.30427803417922 , 1.08655868846497),
(3.30429673329093 , 1.08655930694968),
(3.30432478121775 , 1.08655930174652),
(3.30433278932944 , 1.08655308246640),
(3.30446348355532 , 1.08638330933224)
}'::spoly;

spoly                                                                                             
--------
 {(3.30474723646012 , 1.086004562053),(3.30341855309927 , 1.08577960186707),(3.30341054542378 , 1.08578643990271),(3.30297351563319 , 1.086335
34556428),(3.30357156120003 , 1.0864368395721),(3.30358891855857 , 1.08643995044436),(3.30360894676365 , 1.08644306147078),(3.30361829343581 ,
 1.08644430596871),(3.30362630482521 , 1.08644555030213),(3.30364633346451 , 1.08644866102),(3.30365300940335 , 1.08645052692055),(3.303661020
96957 , 1.08645177113937),(3.30367036769496 , 1.08645363721023),(3.30367837934959 , 1.08645488137174),(3.30368906174976 , 1.08645612569695),(3
.30370107936906 , 1.08645799183673),(3.30370642025712 , 1.08645985750225),(3.30373179124734 , 1.08646358962156),(3.30374514456618 , 1.08646545
561358),(3.30410706158729 , 1.08652886672786),(3.30427803417922 , 1.08655868846497),(3.30429673329093 , 1.08655930694968),(3.30432478121775 , 
1.08655930174652),(3.30433278932944 , 1.0865530824664),(3.30446348355532 , 1.08638330933224)}
(1 row)

the failures:

psql (15.4, server 15.5)
Type "help" for help.

content=> select pg_sphere_version();
 pg_sphere_version 
-------------------
 1.3.1
(1 row)

content=> select '{
(3.30474723646012 , 1.08600456205300),
(3.30341855309927 , 1.08577960186707),
(3.30341054542378 , 1.08578643990271),
(3.30297351563319 , 1.08633534556428),
(3.30357156120003 , 1.08643683957210),
(3.30358891855857 , 1.08643995044436),
(3.30360894676365 , 1.08644306147078),
(3.30361829343581 , 1.08644430596871),
(3.30362630482521 , 1.08644555030213),
(3.30364633346451 , 1.08644866102000),
(3.30365300940335 , 1.08645052692055),
(3.30366102096957 , 1.08645177113937),
(3.30367036769496 , 1.08645363721023),
(3.30367837934959 , 1.08645488137174),
(3.30368906174976 , 1.08645612569695),
(3.30370107936906 , 1.08645799183673),
(3.30370642025712 , 1.08645985750225),
(3.30373179124734 , 1.08646358962156),
(3.30374514456618 , 1.08646545561358),
(3.30410706158729 , 1.08652886672786),
(3.30427803417922 , 1.08655868846497),
(3.30429673329093 , 1.08655930694968),
(3.30432478121775 , 1.08655930174652),
(3.30433278932944 , 1.08655308246640),
(3.30446348355532 , 1.08638330933224)
}'::spoly;
ERROR:  spherepoly_from_array: a line segment overlaps or polygon too large

Same for pgsphere-1.4.1.

The postgresql servers are running in fedora 38 lxd containers using postgresql15-server-15.5 from yum.postgresql.org and I built the pgsphere rpms from source (git clone, checkout tags, make source tarball, use rpmbuild). A wide range of testing of these different versions was good, but I ran into this one that I can't explain.

[vitcpp]

@pdowler There is a severe bug in polygon validation function that was introduced recently some months ago, unfortunately. I've prepared a PR. If you don't mind I will use your polygon data in tests. Thank you very much for reporting!

[pdowler]

thnx @vitcpp - yes, please retain that example for tests