Actually shouldn't it be failures="" errors="%d" then since the test failed to generated the expected test result?

Actually shouldn't it be failures="" errors="%d" then since the test failed to generated the expected test result?

I agree with this in the case of _generate_results() is being called in the presence of problems affecting the own testing code (I think that this is being done in https://github.com/ament/ament_cmake/blob/master/ament_cmake_test/ament_cmake_test/__init__.py#L104). I see more points in the code where _generate_results() is being called and probably they are not related to test code problems.

In that case the current form is valid, in the first scenario you are right, errors="%d" failures="0" should be more close to the specification.

I agree with this in the case of _generate_results() is being called in the presence of problems affecting the own testing code (I think that this is being done in https://github.com/ament/ament_cmake/blob/master/ament_cmake_test/ament_cmake_test/__init__.py#L104). I see more points in the code where _generate_results() is being called and probably they are not related to test code problems.

In that case the current form is valid, in the first scenario you are right, errors="%d" failures="0" should be more close to the specification.

I'll follow this in a new PR with some code changes.