OCT18 challenge CHEFTAK's testcases confirmed to be completely broken

challenge
issue
long-contest
oct18

#1

@admin @vijju123 @mgch

(Just a fair warning, this will be a angry rant. But I really feel like that’s apt in this case.)

(Also, note that this finding very likely means that it is literally impossible to get a non-zero score in this challenge with the current data)

Invalid input

Through quite a few submissions checking the range of stuff I now know that t seems fine. Both Y and Z however are always zero!!! Not only that, both are exactly the string 0.000000 which is exactly the standard float format for 0 in printf. This can only occur if printf is called with the float 0, or if something like an int is given to printf to format as a float. I predict that int is used internally for accuracy and then something like printf("%d %f %f ", t, y/LARGE_FACTOR, z/LARGE_FACTOR); is in the code so that ints are sent to printf to be formatted as float.

How the actual fuck has no-one on the team responsible for this challenge not realized that all the cases generated have Z == Y == "0.000000"? Just looking at the generated data for a split second would raise immediate concerns. How can you say the issue was fixed when the data is this obviously wrong? Have you run any single check? Does it really take a contestant essentially doing black box debugging to find the error the people with the actual source code can’t find? This is absurd.

Both these code snippets runs and does not get NZEC (the division by zero is not encountered):

t = 5
for _ in range(t):
    n = int(input())
    for _ in range(n):
        t,y,z = input().split()
        t = int(t)
        y = float(y)
        
        if z != '0.000000':
            1/0
        print(0)

t = 5
for _ in range(t):
    n = int(input())
    for _ in range(n):
        t,y,z = input().split()
        t = int(t)
        
        if y != '0.000000':
            1/0
        print(0)

[/rant]


#2

No one can’t say anything about this month because this month is full of this kinda contests


#3

also, is there a t (=5) at the top to read in or not? (the number of test cases.) because if i fix the division by zero issue with a try/except clause, but try to read t, i still get a runtime error. however, if i then remove that part and only start by reading in n (=10000), the runtime error disappears.

how is it so difficult to get the formatting correct?? this is a complete joke at this point. no problem setting should be THIS sloppy.


#4

I wish that this should not be fixed.


#5

If what you’re saying is “just end the contest already” I’m completely with you.


#6

@algmyr - what he meant was…if this is fixed…perhaps another extension might be granted to solve cheftak. So well…XDDDDDDD


#7

I forwarded this urgently to @admin . I understand that this long has been frustrating for you, well, the 4 day extension was overall a bad idea. I had talks with @admin to revert but it failed there. But I did make enough of a rant to make sure such a situation doesnt occur next time. Thats all I can say without explicitly going into details of conversations.


#8

What I just don’t get is how someone looked at this and said “this is fixed” when it obviously isn’t fixed. Even a cursory glance at the test cases or writing the simplest of attempts would uncover these problems. Why should I trust the people running this long contest if they couldn’t even do this level of basic testing before releasing it?


#9

I had the pleasure of working side to side with @mgch in april and september long. He does his job responsibly, if such an error occurred this time I will say its more of an exception. Perhaps because he has to look at Snackdown rounds and due to which he committed mistakes in a hurry. But as far as my experience goes,he gives his all in testing problems. Sorry, that might not answer your question precisely,but I had to speak up for him as well. I’d just request to take this long as an exception rather than a habit. I feel he also completely depended on assert statements like you. Thats all :slight_smile:


#10

Any comment about Sept long, Sept CookOff, Sept LTime, and Oct CookOff??


#11

The problem was just removed from the contest. But yes, the t is not present in the data. The data is essentially five runs of:

10000
1 0.000000 0.000000
2 0.000000 0.000000
...
1 0.000000 0.000000
1 0.000000 0.000000