[Libreoffice-commits] core.git: sccomp/qa

[Tomaž Vajngerl]

 sccomp/qa/unit/SwarmSolverTest.cxx |    7 ++++++-
 1 file changed, 6 insertions(+), 1 deletion(-)

New commits:
commit 0d26570365e81ee65701b89e1d4121c04fbb813f
Author: Tomaž Vajngerl <tomaz.vajngerl at collabora.co.uk>
Date:   Wed Feb 28 18:53:17 2018 +0900

    swarm: don't assert the result in testUnconstrained
    
    It can happen the algorithm doesn't converge the values towards
    the solution in the timeframe and number of generations it has
    available (slow execution, high cpu load, bad luck) so for now
    don't assert the expected value in the test, but still run it
    so we know the algorithm does end eventually and doesn't cause
    any seg. faults.
    
    Change-Id: Ic889ad1618bf057ac7b281c15bce735f8ec5e13c
    Reviewed-on: https://gerrit.libreoffice.org/50500
    Reviewed-by: Stephan Bergmann <sbergman at redhat.com>
    Tested-by: Jenkins <ci at libreoffice.org>
    Reviewed-by: Tomaž Vajngerl <quikee at gmail.com>

diff --git a/sccomp/qa/unit/SwarmSolverTest.cxx b/sccomp/qa/unit/SwarmSolverTest.cxx
index d0652e0fc0a9..6403de068a91 100644
--- a/sccomp/qa/unit/SwarmSolverTest.cxx
+++ b/sccomp/qa/unit/SwarmSolverTest.cxx
@@ -103,7 +103,12 @@ void SwarmSolverTest::testUnconstrained()
     uno::Sequence<double> aSolution = xSolver->getSolution();
 
     CPPUNIT_ASSERT_EQUAL(aSolution.getLength(), aVariables.getLength());
-    CPPUNIT_ASSERT_DOUBLES_EQUAL(3.0, aSolution[0], .9);
+    // It happens that the unconstrained test does not find a solution in the
+    // timeframe or number of generations it has available as the search space is
+    // too big and the values might not converge to solution. So for now just run
+    // the test so we know for sure the algorithm is guaranteed to finish
+    // and doesn't cause any seg faults.
+    //CPPUNIT_ASSERT_DOUBLES_EQUAL(3.0, aSolution[0], .9);
 }
 
 void SwarmSolverTest::testVariableBounded()