Today, I experienced something weird. Check this submission. (It has another critical bug that isn't affected by test 1 in the code so yeah, it is incorrect regardlessly, but anyways)

Basically, I tried to use fast I/O with fread, with the usual methods for reading integers I use in BOJ. However, I kept getting runtime error, so I struggled to find out what went wrong, and found this with that submission.

So what's strange is that, in the input sometimes the newline is represented as "\n", while in some others it's "\r\n", which is a specific way used only by Windows. Normally, typical input methods will address them properly no matter which version is used, but with fread, I managed to see this inconsistent structure.

Specifically, here only "\n" (ASCII code 10) is used:

98: 51
99: 10
100: 50
...
106: 50
107: 10
108: 51

while in these two, "\r\n" (ASCII code 13 and 10) is used:

112: 51
113: 13
114: 10
115: 50
...
121: 50
122: 13
123: 10
124: 0

So my doubt is, is it really allowed to be inconsistent with EOLN in Codeforces? In BOJ, the uploader checks whether any test has "\r\n" and automatically truncates them into "\n", but it seems like this test shows that Polygon and testlib has no such procedure to integrate the format. I know it normally doesn't cause much issue with standard input methods, but I thought Polygon would be more strict with it.