PostGIS 2.0 tickets regress winnie failure against geos 3.3.9

[robe]

This might have to do with testing 3.3.9. Only seeing failure on winnie.

 tickets .. failed (diff expected obtained: /c/jenkins/postgis/tmp/2.0.5SVN_pg9.2_geos3.3.9dev_gdal1.9.2w32/test_66_diff)
-----------------------------------------------------------------------------
--- tickets_expected	2013-09-08 16:11:25 -0400
+++ /c/jenkins/postgis/tmp/2.0.5SVN_pg9.2_geos3.3.9dev_gdal1.9.2w32/test_66_out	2013-10-10 08:46:57 -0400
@@ -242,4 +242,4 @@
 #2423|POLYGON((-10 0,-9.2388 3.82683,-7.07107 7.07107,-3.82683 9.2388,0 10,3.82683 9.2388,7.07107 7.07107,9.2388 3.82683,10 0,-10 0))
 #2424|MULTILINESTRING((0 0,10 0,24 3,30 10))
 #2427|POINT(-1 0)
-#2168|5340.762373947|5340.762373947
+#2168|5340.762373948|5340.762373948

[robe]

I don't think it has to do with geos since this is a distance test. I think this is only failing for w32 (9.3). Wondering if the numbers should just be rounded.

[robe]

make that I think just w32 (9.2) the w64 works fine.

[strk]

After looking at #2168 I think we should really change the test to run equality between the first and the second, rather than showing exact numbers.

[strk]

_also_ keeping a test for a rounded number (not too much) would be good, just to be sure the answer isn't completely wrong

[robe]

fixed for 2.1 at r12029

[robe]

fixed for 2.0 at r12030

[robe]

oops I just realized I rounded too much that it wiped out the test I was trying to validate. Have to commit again once I've tested.

[robe]

revised commit of 2.1 at r12031

[strk]

I like the last one (r12031). Backport to 2.0 and close ?

[robe]

yeh and I need to up port to 2.2 as well. I'll do that later today

[robe]

Backported to 2.0 at r12034

[robe]

upported to 2.2 at r12035