You’ll often write a function that works only in some situations. For example, you might write a function that takes as a parameter a number of people who want to eat a pie and returns the percentage of the pie that each person gets to eat. If there are five people, each person gets 20% of the pie; if there are two people, each person gets 50%; if there is one person, that person gets 100%; but if there are zero people, what should the answer be?
Here is an implementation of this function:
| | def pie_percent(n: int) -> int: |
| | """Assuming there are n people who want to eat a pie, return the |
| | percentage of the pie that each person gets to eat. |
| | |
| | >>> pie_percent(5) |
| | 20 |
| | >>> pie_percent(2) |
| | 50 |
| | >>> pie_percent(1) |
| | 100 |
| | """ |
| | |
| | return int(100 / n) |
Reading the code, if someone calls pie_percent(0), then you probably see that this will result in a ZeroDivisionError. There isn’t anything that anyone can do about this situation; there isn’t a sensible answer.
As a programmer, you warn other people about situations that your function isn’t set up to handle by describing your assumptions in a precondition. Here is the same function with a precondition:
| | def pie_percent(n: int) -> int: |
| | """Precondition: n > 0 |
| | |
| | Assuming there are n people who want to eat a pie, return the percentage |
| | of the pie that each person gets to eat. |
| | |
| | >>> pie_percent(5) |
| | 20 |
| | >>> pie_percent(2) |
| | 50 |
| | >>> pie_percent(1) |
| | 100 |
| | """ |
| | |
| | return int(100 / n) |
Whenever you write a function and you’ve assumed something about the parameter values, write a precondition that lets other programmers know your assumptions. If they ignore your warning and call it with invalid values, the fault does not lie with you!