Well, that's how every BigInteger/Decimal class/module in many languages work.

If you want to pass an exact value into their constructors, you gotta pass it in as a string.

I guess he passed in the argument like this and so it didn't work:

Fraction(Decimal(1.85))

Fraction from Decimal gives the same result for 1.85 as Fraction from Float.

???

>>> from decimal import *
>>> from fractions import *
>>> Fraction(Decimal('1.85'))
Fraction(37, 20)

Good pick. The problem comes from Python (same with Ruby) conversion of "float" to "rational". I changed my solution and re-published it to get correct results but I suppress the case (1.85, 60) in order not to invalidate all the already passed solutions. With Python (and Ruby) Fraction(1.85) is (4165829655317709, 2251799813685248) instead of (37, 20)... Thanks for your post!

spoiler flags, please! (visible from the dashboard!)