A scientific approach to debugging

| categories: debugging, work

Recently, a friend got in touch to ask for some help:

When it comes to debugging an issue, I'm able to set a breakpoint and debug into the test to see the error - like a stack trace or whatever - but when it comes to fashioning a fix, the actual issue is often something else. Like the stack trace is a symptom, not a cause. How do I train my brain to figure out where the actual root causes of things are?

I was curious what my team at $currentplace thought, so I forwarded the question to them over Slack. Colleagues from across our engineering teams dropped by to help, and we came up with some notes to pass back to my friend.

However, someone ratted me out to the corporate blog crew. :o) So this post was born:

A scientific approach to debugging

It includes my favourite debugging story, about Maurice Wilkes, which I first came across via Russ Cox.