|
|
|
|
|
by foobarchu
829 days ago
|
|
You don't do a single proof of the entire 10k line program, you do proofs of the many small functions that make it up. If you can use each proof as an axiom going into proving the functions that use it, then each of them becomes easier to prove too. If I have a function that might sometimes mutate a variable off in another part of the program, that's just inherently harder to reason about that one that is idempotent and guaranteed to have no side effects. |
|