i have the very same problem as mdczaplicki and i don't get your answer, csc4pg. can you write down, how the definition of functions should look like? i mean, instead of

def to_roman(self, n):
def from_roman(self, s):

is what exactly supposed to be on that lines? and don't tell me that would be spoiler, it's a manner of understanding the quest, you tell me no algorithm by that...

"Run Suite' writes: [[7, 0, 5, 6, 2, 0, 8, 0, 0], [0, 2, 0, 8, 0, 9, 0, 7, 5], [3, 0, 8, 7, 4, 5, 0, 2, 1], [5, 3, 0, 2, 0, 6, 0, 1, 0], [0, 0, 2, 0, 0, 0, 5, 0, 0], [0, 7, 0, 5, 0, 4, 0, 6, 2], [2, 5, 0, 0, 6, 7, 0, 8, 4], [0, 8, 0, 4, 5, 2, 0, 9, 0], [0, 0, 7, 0, 0, 0, 2, 5, 0]] should equal [[7, 1, 5, 6, 2, 3, 8, 4, 9], [6, 2, 4, 8, 1, 9, 3, 7, 5], [3, 9, 8, 7, 4, 5, 6, 2, 1], [5, 3, 9, 2, 7, 6, 4, 1, 8], [4, 6, 2, 1, 9, 8, 5, 3, 7], [8, 7, 1, 5, 3, 4, 9, 6, 2], [2, 5, 3, 9, 6, 7, 1, 8, 4], [1, 8, 6, 4, 5, 2, 7, 9, 3], [9, 4, 7, 3, 8, 1, 2, 5, 6]]

but if i enter the first matrix into the program at home, it returns the second matrix, correctly. so it looks like the testing mechanism bug to me because why program which solved 20 sudokus including this one should just do nothing with input out of the sudden... am i wrong?

UPDATE: i beg pardon, the testing mechanism is right. now my problem is:

solve(problem ); solve(problem2) does not return the same mixed result as
solve(problem2); solve(problem )

in both cases the solve function properly finishes the first board and do nothing with the second. that's kinda weird to me and i have a feeling i got smarter once i find out that. if anyone have a basic idea of why that's suppose to happen, i will gladly read the comments