Failed unit test: rise::scenario::pathloss::tests::ITUPathlossTest::testRNG

Bug #514471 reported by Jan
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
openWNS RISE
Fix Committed
Critical
Daniel Bültmann

Bug Description

Hi,

after a clean checkout from launchpad, one unittest fails on our systems:

1) test: N4rise8scenario8pathloss5tests15ITUPathlossTestE::testRNG (F) line: 229 framework/rise/src/scenario/pathloss/tests/ITUPathlossTest.cpp
double equality assertion failed
- Expected: 1
- Actual : 0.881695160600658
- Delta : 0.03

After looking at the code in question in framework/rise/src/scenario/pathloss/tests/ITUPathlossTest.cpp it seems to me that it basically checks whether a number of random values from the detail::HashRNG have a mean value close enough to 1. Maybe, on our systems these random numbers come in a different sequence and maybe it was a coincidence that the mean did not deviate much from 1 on your systems.

When changing the stepsize from 0.1 to 0.01 in that test (creates 1000x as many random values) the mean gets closer to 1 but still is not close enough for the test to succeed:

- Expected: 1
- Actual : 0.935649913756681
- Delta : 0.03

So, do you think it is a bug in the simulator, a bug on our system here or was it just a coincidence that it worked OK for you? Or does the test fail for you as well?

Cheers
Jan

Revision history for this message
Maciej Muehleisen (mue-comnets) wrote :

We can config the problem and will provide a fix soon.

Changed in openwns-rise:
importance: Undecided → Critical
status: New → Confirmed
assignee: nobody → Daniel Bültmann (daniel.bueltmann)
Revision history for this message
Daniel Bültmann (daniel.bueltmann) wrote :

Hi Jan,

do you work on a standard 64bit Ubuntu lucid system? We had a problem when migrating to 64bit with the boost hashing function. I wonder if this is the same problem. Can you report your Ubuntu version, boost version and your compiler target architecture?

Regards,
  Daniel

Changed in openwns-rise:
status: Confirmed → Incomplete
Revision history for this message
Maciej Muehleisen (mue-comnets) wrote :

We now use our own Hashing function which is system independent

Changed in openwns-rise:
status: Incomplete → Fix Committed
To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.