I check it by calling different methods in different test cases.
But CodeRunner stop compilation other cases, after the first fail.
So I need to split my task in several tasks.
It is not critical, but not very useful, I think.
My students have to write a class with several methods.
CodeRunner was developed assuming All-Or-Nothing grading would be the norm. Usually compile and runtime errors affect all tests; a result table with the same message repeated over and over again in bright red for each test isn't very user friendly. It can also be very inefficient, e.g. if a program is looping and timing out on each test. So I abort testing after such errors.
However, your case is indeed different. So ... have you considered using a combinator template grader? That gives you total control of the whole process, including generation of the result table.
Richard