Hi Codeforces and problem setters!

While preparing a problem in Polygon, I once again ran into a classic headache when validating tests with incorrect/suboptimal solutions (e.g., greedy approaches, wrong heuristics, or $$$O(NK)$$$ DP instead of $$$O(N \log C)$$$).

The Problem

Imagine a standard situation: a problem setter writes a flawed solution to make sure the test suite catches it.

Here is what happens in practice:

1. On small tests at the beginning of the testset, the code outputs an incorrect result and gets Wrong Answer.
2. On large tests ($$$N = 10^5$$$), the exact same code hangs in an unoptimized loop and gets Time Limit Exceeded before it even gets a chance to output a wrong answer.

Question: Which tag should we assign to this solution in Polygon?

• WRONG_ANSWER — Fails with PackageException, because on large tests the code gets TLE instead of WA.
• TIME_LIMIT_EXCEEDED — Fails with PackageException, because on small tests the code gets WA instead of TLE.
• TL_OR_OK / WA_OR_OK — Not applicable, because the solution should not pass the full testset anyway.
• FAILED — Fails with PackageException, because Polygon expects a runtime or logic error, but the process gets killed by TLE on larger tests.

Current Workarounds (And Why They Suck)

Right now, problem setters have to invent ugly crutches just to satisfy Polygon's strict tag system:

1. Adding artificial cutoffs like if (n > 1000) return 0; so the greedy code terminates instantly and gets a clean WA without timing out.
2. Manually crafting micro counter-tests and shuffling them to the very top of the test list.
3. Simply deleting these flawed solutions from the repository, which directly lowers the quality of testsuite validation.

We spend time micro-optimizing intentionally wrong solutions just to appease the invocation checker, which feels counterproductive.

Proposed Solution

Add a WA or TLE tag (or a broader INCORRECT / ANY_ERROR tag) to Polygon.

Proposed Tag Logic:

A solution passes verification if it fails on the testset with either WA, TLE, ML, or RE, as long as it is rejected by the testsuite and does not get an overall OK.

This small addition would save problem setters a ton of time and make task development in Polygon much smoother.

What do you think?

UPDATE: Turns out I was just sleep-deprived. The Incorrect tag does exactly what I needed. Thanks for reading!