Sign in to watch this lesson
Algorithms - Capstone · 4:14
Stress Testing and Debugging Under Pressure
When a solution passes the samples but fails hidden tests, stop staring at the code and build three small programs: a brute force you trust, a random generator and a loop that compares them. We run a real stress loop on a Mac until a four-number input breaks a fast solution, shrink it to two numbers, and find an int overflow that the undefined-behaviour sanitizer confirms in one line. Then we line up the usual suspects: off-by-one reads, unreset state, deep recursion and endl flushing, with the flags that catch them.
You will learn
- Build a stress test from a brute-force solution, a seeded random generator and a loop that stops at the first difference
- Design generators that find bugs: small n so failures are readable, values that reach the whole allowed range
- Shrink a failing input by deleting elements while it still fails, then read the bug off the minimal case
- Recognize the usual suspects (int overflow, off-by-one, uninitialized or unreset state, recursion depth, endl flushing) and catch them with asserts, -fsanitize=address,undefined and the standard library's debug checks