Hacker News Reader

Comment by jacquesm | Hacker News Reader

The first three have solved just about every problem that I've ever encountered in software, even in systems that I had absolutely no clue about (which made the first two quite natural ;) ).

Debugging and troubleshooting isn't art or magic, it's just being very boringly methodical and building from the ground up. A single assumption and you're off to a dead end, a single thing taken on hearsay and you will repeat what other people already did, failure to challenge and check will no doubt make you miss something that it crucial to solving the problem.

View on HN

Theres a quote I read somewhere (paraphrased) that the scientific method is the slowest method of debugging but it always eventually succeeds. Most bugs are fixable most quickly by intuition, or making a ton of assumptions about where the problem is likely to be. Its usually at my lowest point I decide to resort to “science” and hours later I finally find the problem.

Agreed. At the bottom of any long-to-debug issue I've encountered is some assumption I absolutely knew to be beyond suspicion that turned out to be wrong.

(That said, it's easy to say in retrospect. In the moment there is always judgement about what to verify first, and of course its always the last thing you check that turns out to be the problem)

This is the harsh thing about computers: they mercilessly point out your character flaws.

Well it would be silly to keep checking once you’ve already found the problem, surely? /s

I don’t understand the “/s” sarcasm tags, but there can be cases where you do want to keep checking.

Some problems have a single root cause, some have multiple independent root causes, and some issues present when root causes have an interaction type effect.

You have to both find AND recognize that you've found.

I have noticed that when a Stack Overflow question says "the problem can't possibly be X, because <something>", that is very often exactly where the problem is (and <something> is a misunderstanding).

Hehe, the 'stackoverflow blindspot'.

Probably you could extend your theory to 'first google result for question 'x''.

Reply on news.ycombinator.com