On Mon, Mar 02, 2015 at 04:51:35PM +0100, 'Klaus Aehlig' via ganeti-devel wrote:
When updating statistics to obtain new standard deviations,
rounding errors do happen. Testing for an accuracy of 1e-10
is still an order of magnitude better than our limit of 1e-9
for taking action.
Signed-off-by: Klaus Aehlig <[email protected]>
---
test/hs/Test/Ganeti/Utils.hs | 2 +-
1 file changed, 1 insertion(+), 1 deletion(-)
diff --git a/test/hs/Test/Ganeti/Utils.hs b/test/hs/Test/Ganeti/Utils.hs
index b99c909..844c215 100644
--- a/test/hs/Test/Ganeti/Utils.hs
+++ b/test/hs/Test/Ganeti/Utils.hs
@@ -342,7 +342,7 @@ prop_stddev_update =
direct = stdDev modified
in printTestCase ("Value computed by update " ++ show with_update
++ " differs too much from correct value " ++ show direct)
- (abs (with_update - direct) < 1e-12)
+ (abs (with_update - direct) < 1e-10)
-- | Test list for the Utils module.
testSuite "Utils"
--
2.2.0.rc0.207.ga3a616c
LGTM
