Add a (broken, disabled) test for non 100% node cache hit ratio

apmon · apmon · commit 20a4a1ddb111 · 2013-08-30T11:19:10.000-06:00
To test the node retrieval from database during initial import, add
a test case for which not 100% of nodes fit into memory cache.

The minimum cache size of 1MB is allready to big in optimized mode.
In dense mode, things are sufficiently inefficient that even liechtenstein
needs more than 1MB of node cache.

However, --cache-strategy=dense appears broken at the moment. It gives the
wrong geometries, likely due to wrongly retrieved nodes. This needs fixing
before we can activate this test.
diff --git a/tests/regression-test.py b/tests/regression-test.py
@@ -186,6 +186,8 @@ def __init__(self):
                                               "testTwo")))
         self.addTest(BasicSlimTestCase("basic case", [], [0,1,2,3,13, 91, 92],[6,7,8,9]))
         self.addTest(BasicSlimTestCase("Parallel processing", ["--number-processes", "8", "-C100"], [0,1,2,3,13,91,92],[6,7,8,9]))
+        #It looks like --cache-strategy=dense is broken, as it it produces wrong geometries.
+        #self.addTest(BasicSlimTestCase("Parallel processing with non 100% node-cache", ["--number-processes", "8", "-C1", "--cache-strategy=dense"], [0,1,2,3,13,91,92],[6,7,8,9]))
         # Failes to do correct error checking. This needs fixing in osm2pgsql
         # self.addTest(BasicSlimTestCase("Parallel processing with failing database conneciton (connection limit exceeded)", ["--number-processes", "32", "-C100"], [0,1,2,3],[6,7,8,9]))
         # Counts are expected to be different in hstore, needs adjusted tests
