Oracle GIS update

145 views
Skip to first unread message

Jani Tiainen

unread,
Nov 5, 2015, 3:21:23 PM11/5/15
to django-d...@googlegroups.com
Hi, I finally had time to get back on Oracle GIS issues.

I ran test suite against 1.9.x and 7 tests fails.

3 tests are most probably due different algorithms used to calculate geographical distance and areas. My proposal for fix is to use backend spesific values.

Failing testcases are:
======================================================================
FAIL: test_distance_function (gis_tests.geogapp.tests.GeographyFunctionTests)
AssertionError: 4899.67677194225 != 4891.2 within 2 places
======================================================================
FAIL: test_geography_area (gis_tests.geogapp.tests.GeographyFunctionTests)
AssertionError: 5439100.13586914 != 5439100.95415646 within 5 places
======================================================================
FAIL: test06_geography_area (gis_tests.geogapp.tests.GeographyTest)
AssertionError: 5439100.13586914 != 5439100.95415646 within 5 places
======================================================================



4 tests are unknown failures that require further investigation.
Failing testcases are:
======================================================================
FAIL: test_difference (gis_tests.geoapp.test_functions.GISFunctionsTests)
======================================================================
FAIL: test_difference_mixed_srid ======================================================================
FAIL: test_intersection (gis_tests.geoapp.test_functions.GISFunctionsTests)
======================================================================
FAIL: test_sym_difference (gis_tests.geoapp.test_functions.GISFunctionsTests)
======================================================================

--
Jani Tiainen


Marc Tamlyn

unread,
Nov 10, 2015, 1:54:42 PM11/10/15
to django-d...@googlegroups.com

Thanks for getting it running at all!

It looks to me like those tests just need to be less accurate for Oracle, good plan.

Did you have any specific changes to make to get it to run at all? If so we should include those soon, irrespective of whether the fixes for these other tests make it into 1.9 final.

Marc

--
You received this message because you are subscribed to the Google Groups "Django developers (Contributions to Django itself)" group.
To unsubscribe from this group and stop receiving emails from it, send an email to django-develop...@googlegroups.com.
To post to this group, send email to django-d...@googlegroups.com.
Visit this group at http://groups.google.com/group/django-developers.
To view this discussion on the web visit https://groups.google.com/d/msgid/django-developers/CAHn91ofJ0QQjKJ9KZXwBMLhyD59AN6sxq%3D_JtNz6xeuujH8tEw%40mail.gmail.com.
For more options, visit https://groups.google.com/d/optout.

Tim Graham

unread,
Nov 10, 2015, 2:23:17 PM11/10/15
to Django developers (Contributions to Django itself)
The changes to get it running were merged in September. Absent a proper solution for those 6 failures, I agree it would be a good idea to mark them expectedFailure so we can get the build running on CI before 1.9 final.

Jani Tiainen

unread,
Nov 11, 2015, 6:12:34 AM11/11/15
to django-d...@googlegroups.com
Well I fix those algorithm based errors, and do deeper analysis for those yet unknown ones. If I can't figure out why that happens I create ticket for each and mark them expected failures.

Jani Tiainen

unread,
Nov 19, 2015, 6:47:49 AM11/19/15
to django-d...@googlegroups.com
As noted by some of you managed to get it done.

Most problematic thing seems to be "Texas" geometry which causes problems in Oracle and sometimes with SpatiaLite as well.

I'll try to investigate if that data could be improved to create more consistent test results.


On 10.11.2015 21:23, Tim Graham wrote:
Reply all
Reply to author
Forward
0 new messages