Discussion

Ask HN: How to avoid giving a scathing code review?

For the past several hours I've been struggling to unravel a pull request from a newly hired coworker. It was apparent from a quick glance that it needed some work but I underestimated how bad it was. The main logic is in a monolithic ~400LOC function (in a dynamic language, not verbose Java or C) consisting of a deeply nested if/else spaghetti code. I wouldn't dare touching it with a 10 foot pole without tests but fortunately there are some and the original test coverage isn't bad. So I've cut it down to almost half the size so far but there's still way to go until it reaches a somewhat maintainable state. I've also found and fixed a few bugs along the way so strict refactoring is not even a goal at this point.

So I'm wondering how to go about this. A few code review comments here and there won't cut it. If I was to be brutally honest, I'd outright reject the PR and tell him to rewrite it from scratch in a clean way but that's probably not the ideal response. How do you deal with situations like this?

66 points reinhardt 11 years ago 67 comments

Discussion

Live from the official Hacker News API

A modern reader for Hacker News · SvelteKit + Skeleton