You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, if we have a TLE submission, verifyproblem expects that it should always TLE or AC (but TLE on at least one input). It would be natural and nice if verifyproblem allowed a TLE submission to give WA on some inputs, as long as it gives TLE on at least one input.
The intent here is that if we wish to accept an $O(n^2)$ solution but $O(n^3)$ should TLE, then we could make a simple program which just reads the input, ignores the problem, and loops n^3 times. However, this will give the WA error described above.
@austrin points out (via email) that as we move to setting time limits by running submissions through Kattis judge rather than within the problemtools itself, this issue becomes trickier to deal with.
The text was updated successfully, but these errors were encountered:
A note: this should cut both ways: a WA solution should be allowed to give TLE on some test cases, as long as it gives WA on at least one test case. (And the same thing goes for RTE submissions.)
Note that there is a small but non-zero risk that allowing this will allow some human errors to slip through the cracks. For example, if the input format changes, it might cause a WA solution to do undefined things, sometimes giving RTE and sometimes giving WA; with the current behaviour that would be detected and fixed. There may be merit in having some opt-in mechanism (like a magic comment in the source) to indicate which outcomes (other than AC and the goal) are acceptable.
Currently, if we have a TLE submission, verifyproblem expects that it should always TLE or AC (but TLE on at least one input). It would be natural and nice if verifyproblem allowed a TLE submission to give WA on some inputs, as long as it gives TLE on at least one input.
The intent here is that if we wish to accept an$O(n^2)$ solution but $O(n^3)$ should TLE, then we could make a simple program which just reads the input, ignores the problem, and loops n^3 times. However, this will give the WA error described above.
@austrin points out (via email) that as we move to setting time limits by running submissions through Kattis judge rather than within the problemtools itself, this issue becomes trickier to deal with.
The text was updated successfully, but these errors were encountered: