Ask HN: Longest you've spent debugging a single issue? | Hacker News Reader

Ask HN: Longest you've spent debugging a single issue?

6 pointsdavismwfl10 comments

Along the lines of this other HN posting: https://news.ycombinator.com/item?id=11729806

What is the longest you have spent in any code base, debugging a single issue. What was it?

I personally spent approximately 15 days finding and fixing a race condition in a custom multiprocessor enabled 3d rendering engine, that I was not part of writing. It was reasonably early in my career but I feel would still take many back to back days to find a solution today. Difference being I know the problem domain a lot better now to know how to break up the problem and where to look for issues.

In the last year, I think the longest is about 3 straight days on an embedded project tracking down bad values getting set seemingly at random (no JTAG etc).

View on HN

Two months. My first time programming .NET and learning on the job. I was creating an application to import data into SalesLogix using their API. It crashed whenever there was a single quote in a particular field. Not knowing much of anything at the time, I assumed it was my own code, so I hammered away at it. I raised the issue many times and never got help, so I just worked around it, knowing it would be a showstopper in production. One day, a co-worker casually mentions to me that I should use the latest version of their API. Sure enough, it worked.

Fuck SalesLogix, btw.

Never used their API, that bad eh?

BTDT in just working around something because of time/cost constraints, but I have that personality that wants to understand and find the root cause even if we build a workaround. That is a good and bad trait, not everyone wants to pay for you to find the issue just cause you want to.

> Never used their API, that bad eh?

It was that bad back in 2003. Knowing what I know now, I would have pressed for a bug report, but it was/is one of those software packages where only a trained/certified person could contact support.

Back then, I knew 0. I had a computer science degree, some hobby programming, but it was nothing compared to real world programming.

At my current job, 2 years, we have a bug that existed before I joined. However, I was able to provide an alternate solution to it. Its hard to quantify the time spent on it because it was always around. When it became a main focus it probably took me 2 weeks to get an alternate solution.

I think this probably happens a lot of the time. Problems exist and existing team members give up on it until someone just says screw it, lets just rewrite that part of the code or someone new comes along and suggests something different.

Focused solely on one bug - probably about 1 week - which was finding a bug inside Apache Thrifts RPC protocol.

But I find you rarely spend time focused on one bug - not just for practical reasons, but this is usually not the best way to fix difficult bugs. If you focus directly on it you're in "alert problem solving mode" rather than "quiet reflective mode" and insight and creativity only come in quiet reflective mode.

I've solved so many bugs while playing minecraft, or standing in the shower, or walking to the train.

If only I could convince my boss that when I'm playing minecraft - I'm only distracting my immediate consciousness (is there a word for this) while my mind is actually reasoning about the bug / rubber ducking

I guess hobby / less experienced programmers dont count in this answer.

My way of phrasing the question was: after how long should I quit debugging if unsucessful?

My answer to your question: a day or two, and not talking about advanced programming.

I havent read books about debugging yet but I started to track reasons for why things go wrong and I'm perplexed. Answering this question makes me feel like I tried to reinvent the wheel. That will determine me to read a book about debugging. Right now I want to find out all the reasons why bugs occur. I dont know much about other programmers, this is an interesting topic to read.

Almost two years of my free time, trying to get node-gyp to install an npm package properly. I've been working on the same side-project for 5+ years now. One alpha, no release yet :(

damn 5 years looks too long

do you work little on it or just expand the features ?

It's ambitous, and when I started, I wasn't very good :)

Reply on news.ycombinator.com