[postgis-devel] PostGIS 1.5 regression with GEOS 3.0.3

[Mark Cave-Ayland]

Hi folks,

I'm seeing the following regression with GEOS 3.0.3 with SVN trunk, but 
not with GEOS 3.1.1 or 3.2.0:

*** tickets_expected    Tue Jan 12 10:10:00 2010
--- /tmp/pgis_reg_7461/test_45_out      Tue Jan 12 10:38:45 2010
***************
*** 1,4 ****
! #2|POLYGON((1 1,1 2,2 2,3 2,3 1,2 1,1 1))
   #11|0
   #21|SRID=31293;POINTM(6220.13 5337367.145 4566)
   t
--- 1,4 ----
! #2|POLYGON((2 2,3 2,3 1,2 1,1 1,1 2,2 2))
   #11|0
   #21|SRID=31293;POINTM(6220.13 5337367.145 4566)
   t

The SQL that produces this appears to be:

SELECT '#2', ST_AsText(ST_Union(g)) FROM
( VALUES
('SRID=4326;MULTIPOLYGON(((1 1,1 2,2 2,2 1,1 1)))'),
('SRID=4326;MULTIPOLYGON(((2 1,3 1,3 2,2 2,2 1)))')
) AS v(g);


This looks like it is just the order that has changed rather than the 
actual result. Should we consider bumping the minimum GEOS version to 
3.1 for PostGIS 1.5 just to help keep things clean? As a plus, we get to 
remove lots more #ifdef cruft...


ATB,

Mark.

-- 
Mark Cave-Ayland - Senior Technical Architect
PostgreSQL - PostGIS
Sirius Corporation plc - control through freedom
http://www.siriusit.co.uk
t: +44 870 608 0063

Sirius Labs: http://www.siriusit.co.uk/labs