Write Test Case for ACM

When resolving a ACM problem, it's not a good idea to simply write a compilable algorithm and submit to the ACM system in the hope of getting AC (accept). This way, we re relying on the system to provide some secret test cases to validate the correctiness of the algorithm. And we lose two things:
1. A good chance to train yourself thinking about things completely.
2. Efficiency, since it will take you more submit to get pass all test cases.

So, it's important to design our own set of test cases. Here are some basic idea

Sample test case provided by the problem

Boundary test case

Random test case

Test case for large scale

Trick test case


Tip:
We can call the algorithm twice and compare their output to make sure we don't forget to initialize a variable on the stack. Consider the example below, if we run the test case to call the algorithm twice, we can find out the annoying mistake. It's a good way to help use avoid some trivial mistakes.

 1 #include "iostream"
 2 
 3 using namespace std;
 4 
 5 int foo()
 6 {
 7   int z; // variable z is declared without initialization.
 8   int a = 13;
 9   int r = a + z;
10   z = 35;
11   return r;
12 }
13 
14 int main()
15 {
16   int r1 = foo();
17   int r2 = foo();
17   assert(r1 == r2);
19 }

How to write test case
In practice, I'm using google's gtest framework to perform my unit test.