TL;DR: I built stress.ojcloud.net — a free online stress tester for competitive programmers. Paste your main, brute, and gen into three tabs, hit Run, and it streams per-test verdicts and stops at the first counterexample with Input / Main Output / Brute Output side-by-side. No signup, no install, 8 languages (C++ 03/11/14/17/20, C, Python/PyPy, Java, Kotlin, Go, Rust).
Hello, Codeforces!
If you've been doing CP for more than a week, you already know the routine. You submit, you get Wrong answer on pretest 2, the test is hidden, and your solution is 300 lines long. You open a terminal, mkdir stress, write the same while true; do ./gen > in; ./a.out < in > out1; ./brute < in > out2; diff out1 out2 || break; done you've written a hundred times, realize you forgot to seed the generator, rewrite it, forget to recompile brute, finally see a diff, and now you're squinting at 2000 random numbers trying to minimize the counterexample by hand.
We all do it. tourist does it, Um_nik does it, I do it. Stress testing is probably the single most valuable debugging skill in CP, and yet every single one of us has hand-rolled the same bash script a hundred times.
The friction with the bash approach
- Boilerplate every single time: compile three files, seed loop, diff, break.
- No UI. You can't see which test is running or how far along you are.
- Switching languages (Python brute, C++ main) means rewriting the runner.
- No history. Close the terminal, lose everything.
- Sharing a counterexample with a teammate means pasting 3 files into Discord.
- Forget to
-O2the brute and it takes 40 seconds per test. Forget to seed the RNG and every test is identical.
I got tired of this, so I built something.
What stress.ojcloud.net does
- 8 languages: C++ 03/11/14/17/20, C, Python 3, PyPy 3, Java, Kotlin, Go, Rust. Switch from a dropdown; templates auto-load.
- 5 built-in generator snippets: Array, Tree, Graph, String, Grid — one click drops working code into the
gentab. Each snippet uses arandom(lo, hi)helper and a seed fromargv[1], so runs are reproducible. - Real-time verdicts: the right panel streams
Test 1 → Test 2 → …live. Stops the moment it finds a mismatch — no waiting for the remaining 900 tests. - Counterexample card: shows
Input,Main Output,Brute Outputside-by-side with one-click copy on each. - Configurable tests & timeout: default 200 tests / 2s per test, tune to taste.
- Browser-resumable session: close the tab mid-run, reopen it, your code and last result are still there.
- 7-language UI: English / Tiếng Việt / Русский / 简体中文 / 日本語 / 한국어 / Português.
- Free. No signup. No account. No email. Just open the page and paste.
Demo: finding a real bug in ~30 seconds
Let's do the most classic bug in CP — Kadane's on max subarray sum, with best = 0 instead of LLONG_MIN. Invisible on positive inputs, fatal on all-negative ones.
main (buggy O(n) Kadane):
#include <bits/stdc++.h>
using namespace std;
int main() {
int n; cin >> n;
vector<long long> a(n);
for (auto& x : a) cin >> x;
long long best = 0, cur = 0; // BUG: should be LLONG_MIN
for (long long x : a) {
cur = max(x, cur + x);
best = max(best, cur);
}
cout << best << "\n";
}
brute (O(n²), definitely correct):
#include <bits/stdc++.h>
using namespace std;
int main() {
int n; cin >> n;
vector<long long> a(n);
for (auto& x : a) cin >> x;
long long best = LLONG_MIN;
for (int i = 0; i < n; i++) {
long long s = 0;
for (int j = i; j < n; j++) { s += a[j]; best = max(best, s); }
}
cout << best << "\n";
}
gen: click the Array snippet, tweak the range to random(1, 8) and random(-10, 10) so negatives actually appear. Hit Run.
Within a few tests the stream stops on:
Input Main Output Brute Output
2 0 -3
-3 -5
There's the bug. All-negative array, buggy Kadane returns 0, correct answer is -3. Fix the init to LLONG_MIN, re-run, green "All 200 tests passed". Total time: under a minute.
Try it
stress.ojcloud.net — open the page, paste 3 files, hit Run. That's it. No signup flow to click through.
If you prefer a walkthrough, here's a 30-second video covering the exact Kadane example above: youtu.be/XI1HNegwFuI.
Feedback / bug reports / feature requests
I'd love to hear what breaks, what's missing, and what languages/snippets you'd want next. Join the Discord: discord.gg/pg52FspEns — share counterexamples, report bugs, argue about generators.
Thanks
Huge thanks to MikeMirzayanov — the checker design for OJCloud's main judge was shaped heavily by testlib.h, and the same philosophy (strict, readable, reproducible) carried over into this stress runner. And thanks to the countless CF blogs over the years that taught everyone the gen / main / brute / diff pattern in the first place — this tool is just that pattern with a UI bolted on.
Happy stressing.
