Opened 3 years ago

Closed 3 days ago

#694 closed defect (fixed)

PreparedPoint::intersects gives different results from Point::intersects

Reported by: strk Owned by: geos-devel@…
Priority: blocker Milestone: 3.6.2
Component: Default Version: 3.4.2
Severity: Unassigned Keywords:
Cc: mtnclimb@…

Description

A bug in PreparedPoint::intersects signature prevented it from being really prepared. The bug was fixed by r3988, which triggered no failures in the GEOS testsuite but triggered a failure in the PostGIS testsuite: http://trac.osgeo.org/postgis/ticket/2768

This makes me think that PreparedPoint::intersects behaves differentely from Point::intersects, which is the one we were hitting while the bug was present.

Change History (8)

comment:1 Changed 3 years ago by strk

I've reverted the signature fix with r3989, but this issue should be fixed or it makes no sense to have an unused implementation for PreparedPoint::intersects.

An automated testcase is needed in GEOS.

comment:2 Changed 3 years ago by strk

Milestone: 3.5.03.4.3
Version: svn-trunk3.4.2

Interesting enough, it looks like tests for prepared predicates are completely missing, see http://trac.osgeo.org/geos/browser/tags/3.4.2/tests/xmltester/tests/general/MISSING

Sounds about time to port those tests!

Assuming the bug is also in 3.4.3 (and would probably also be in 3.3.9)

comment:3 Changed 3 years ago by mloskot

See also #591 and r4038

comment:4 Changed 9 months ago by strk

Milestone: 3.4.33.6.1

Ticket retargeted after milestone deleted

comment:5 Changed 5 months ago by strk

Milestone: 3.6.13.6.2

Ticket retargeted after milestone closed

comment:6 Changed 3 days ago by Sandro Santilli <strk@…>

In 63410b1/git:

Add support for testing prepared geometries operations in XMLTester

Add 3 more tests from JTS testsuite now that they can be run
See #694

comment:7 Changed 3 days ago by Sandro Santilli <strk@…>

In a12a5ec/git:

Add support for testing prepared geometries operations in XMLTester

Add 3 more tests from JTS testsuite now that they can be run
See #694 for 3.6 branch

comment:8 Changed 3 days ago by strk

Resolution: fixed
Status: newclosed

It looks like r3988 was re-committed again as r4078 at some point after this ticket was opened. And now that we have tests checking after the problem (and clearly PostGIS regresses fine too) I think this ticket can be closed.

Note: See TracTickets for help on using tickets.