Rules of the Huawei Codecraft Marathon
Rules for solutions verification and evaluation
At the beginning of the contest participants receive the task condition. During the contest participants can send the task solution; if the problem statement does not explicitly state the contrary, the number of attempts for participant is limited to 2000.
The solution to the problem is code written in one of the installed languages.
If the specific round rules do not expressly indicate special order, tests to problem of the round are divided into three groups.
- "open" tests with answers for the case when the correct answer is known can be downloaded from the test system interface; all submissions are checked by this test immediately after sending.
- "online" tests are not available to participants; all submissions are checked by this test immediately after sending.
- "final" tests are not available to participants; all submissions are checked by this test finally after time for submitting is expired.
The solution for each test is evaluated by integer score from 0 to N_max (value N_max is the maximum score for the test and it is specified in the problem statement). The current result of verification of this solution is equal to the sum of all test scores from groups of "online" and "final". The final result of the participant’s solution testing is equal to the sum of scores of all three groups of tests.
After sending the task, the solution is checked on "open" and "online" test groups and the current result on the solution is communicated to participants. Participants can check the current results table, which is sorted by decreasing maximum values of the current result for all solutions sent. At the end of the contest time all submitted solutions are tested on all three test groups (i.e. test "final" is added). The Final result is calculated after which the final results table is constructed, which is sorted by decreasing maximum values of the current result for all sent solutions. In case of equal scores the team which have sent corresponding result earlier will be rated higher.
In case of solution verification and evaluation in special order it will be described in the problem statement; however the division between "online" and "final" part is retained (the transition can be implemented not only by the addition of tests, but, for example, replacing the checker for a more precise checker with the rating scale).