Debugging: Tracebacks and a Real Fix

Bugs are normal. The skill is finding them calmly, and Python hands you a map: the traceback.

Read tracebacks bottom-up

When code crashes, Python prints a traceback. The last line names the error type and message; that is the most useful part. The lines above show the call chain that led there, with file names and line numbers.

Traceback (most recent call last):
  File "app.py", line 12, in <module>
    print(total / count)
ZeroDivisionError: division by zero

Read the bottom first: a ZeroDivisionError on the line total / count tells you count was 0. Now you know exactly where and why.

Bisect to localize

If you cannot see the cause, narrow it down. Drop a few print() calls (or asserts) to check values at the halfway point of the logic. Whichever half is wrong, bisect again. A handful of prints usually pins the bug fast:

def discount(price, percent):
    print("price:", price, "percent:", percent)   # is the input what I expect?
    factor = percent / 100
    print("factor:", factor)                       # is the math right?
    return price - price * factor

assert your assumptions

An assert states what you believe is true at a point in the code. If the belief is wrong, it fails loudly right there instead of corrupting results silently: assert count > 0, "count should never be zero here".

Here is a function with a subtle bug. It is meant to return the largest number in a list, but it returns the smallest. Read it, find the flaw in the comparison, and fix it.

# buggy
def find_max(numbers):
    best = numbers[0]
    for n in numbers:
        if n < best:      # wrong direction
            best = n
    return best