This is 100% claude code generated, but I believe the problem is real (I had an empty program that pass the tests).

Problem 23 (A=B): All test inputs are multiples of 100, causing checker to skip all validation

Description

Problem 23's custom checker (check.cpp) has an early-exit path at line 351:

void check_Tid(){
    test_gen::Tid=inf.readInteger();
    if (test_gen::Tid%100==0) quitf(_ok,"Test 0.");
}

When Tid % 100 == 0, the checker immediately returns _ok (accepted / full score) without generating any test strings or running the contestant's A=B program.

All three test inputs in testdata/ trigger this path:

This means any submission, including an empty program that produces no output, receives a perfect score of 100.

Reproduction

Submit the following trivial C++ program for problem 23:

#include <bits/stdc++.h>
using namespace std;
int main() {
    return 0;
}

It scores 100/100 on all three test cases.

Root cause

The checker's test generator maps Tid % 100 to a generator function via vf[data_id]. Generator vf[0] is an intentionally empty lambda (produces zero test cases). Valid generators are at indices 1–16+. The early exit if (test_gen::Tid%100==0) quitf(_ok,"Test 0.") was likely intended as a sentinel for the empty-generator case, but all shipped test inputs happen to be multiples of 100, so no test case ever exercises the actual checker logic.

Suggested fix

Replace the test inputs with Tid values where Tid % 100 falls in the range 1–20 (mapping to actual non-empty test generators). For example:

• 1.in: 1 (uses generator vf[1] — hand-crafted cases)
• 2.in: 102 (uses generator vf[2] — random short strings, seed=1)
• 3.in: 206 (uses generator vf[6] — long strings, seed=2)