DiscreteTrajectoryBase push_back unsafe about control-toolbox HOT 5 OPEN

In an effort of helping resolving this issue. I noticed that there is Test for DiscreteArray and DiscreteTrajectory. But there is no Test for DiscreteTrajectoryBase.
Is this the reason that this bug was not discovered before? Should DiscreteTrajectoryBaseTest.cpp be created?

from control-toolbox.

from control-toolbox.

Thank you @markusgft . I was creating the unit test for DiscreteTrajectoryBase, and realized that ct_core/test/DiscreteTrajectoryTest.cpp is actually testing DiscreteTrajectoryBase. Because

    // trajectory containers with zoh and linear interpolation
    StateTrajectory<2> stateTrajectory_zoh(timeArray, stateArray, InterpolationType::ZOH);
    StateTrajectory<2> stateTrajectory_lin(timeArray, stateArray, InterpolationType::LIN);

The StateTrajectory is a derived class from DiscreteTrajectoryBase.
My question, should DiscreteTrajectoryTest.cpp be renamed as DiscreteTrajectoryBaseTest.cpp?
Thank you.

from control-toolbox.

I'm about to create a test case to produce a situation of this bug, and I realized that tpl::TimeArray<SCALAR> has a length of a minimum of 2. It doesn't seem possible that time could be empty. But I believe from the unit test point of view, the unit shall cover that. Could you please clarify? Thanks.

from control-toolbox.

Thanks for raising the issue. I believe that someone needs to take a more detailed look and check what is really going on. I currently have no time to reproduce the failures, unfortunately.

from control-toolbox.