I think that expected and actual lists were not matched in the test case in the c language. they might be good to swap each other.

C++ version generates warnings.

hello there! Can you explain to me how did you come out with that pattern ? (i mean the one stored in the array solver).

Right. I was not defensive enough when I was writing my solution ;-)

Do you really think this is an issue?
void solve(const char* input, char* result) just tells you where to store your result.
It's not unusual for output parameters to be uninitialized, e.g. int int1; if (scanf("%d", &int1) == 1) // ...

Added it myself.

C version: the memory content inside result is not reset during the random tests, which means we have to reset them ourselves.

I just fixed the issues but I have an additional suggestion for the C version.
The current error messages are not that helpful: "The expression test_if_equal(result1, expected_result1) is false."
Why not this:

#define do_test(n) solve(input##n, result##n);\
    cr_assert(test_if_equal(result##n, expected_result##n), \
        "expected: {%d, %d, %d, %d, %d, %d, %d, %d, %d}\n" \
        "actual:   {%d, %d, %d, %d, %d, %d, %d, %d, %d}", \
        expected_result##n[0], expected_result##n[1], expected_result##n[2], \
        expected_result##n[3], expected_result##n[4], expected_result##n[5], \
        expected_result##n[6], expected_result##n[7], expected_result##n[8], \
        result##n[0], result##n[1], result##n[2], \
        result##n[3], result##n[4], result##n[5], \
        result##n[6], result##n[7], result##n[8])

or put some of that into its own function and put it into the "Preloaded" section:

void solve (char* input, char* result);
char test_if_equal(char* input1, char* input2);
void test_solve(char* input, char *expected) {
    char actual[9];
    solve(input, actual);
    cr_assert(test_if_equal(actual, expected),
        "expected: {%d, %d, %d, %d, %d, %d, %d, %d, %d}\n"
        "actual:   {%d, %d, %d, %d, %d, %d, %d, %d, %d}",
        actual[0], actual[1], actual[2],
        actual[3], actual[4], actual[5],
        actual[6], actual[7], actual[8],
        expected[0], expected[1], expected[2],
        expected[3], expected[4], expected[5],
        expected[6], expected[7], expected[8]);
}

In the sample and final tests do_test can then be reduced to #define do_test(n) test_solve(input##n, expected_result##n).

Fixed.

Fixed.

Fixed.

using namespace std;

typedef std::bitset<9> state;

I don't think you really need that using namespace std; on the top ;-)

C

error: called object type 'int' is not a function or function pointer
    do_test(1);
    ^~~~~~~~~~
/home/codewarrior/fixture.c:27:25: note: expanded from macro 'do_test'
#define do_test(n) solve(input##n, result##n);\
                   ~~~~~^

Sample tests missing

• #include <criterion/criterion.h>
• definition of test_if_equal

C++

error: implicit instantiation of undefined template 'std::__1::array<std::__1::bitset<9>, 9>'
  std::array<state, 9> solver = {0x14E, 0x017, 0x163, 0x059, 0x0BA,

error: expected parameter declarator
    std::uniform_int_distribution<int> distribution(0,511);
    

Includes should be done in test code without relying on users' code