[phodge]

> I realized I was spending about a quarter of the total time implementing the feature tracking down which functions the bugs were in!

This was a huge problem for me back when I was doing web development and had to support IE6. I write hundreds of lines of code which worked fine in Firefox, but then it would crash in IE with no descriptive error message and I would spend an eternity gradually commenting out more and more of my changes until I found the offending line through a tedious manual brute force bisect/test process. I was able to increase my productivity by testing every individual little change in IE as I made it, so that when I got some vague error, at least I knew there were only 10 lines of code where it could possibly be originating from.

However, these days with all the linters, typecheckers, and good error tracebacks in browsers, I'm much faster if I leave all my testing for the end. In fact, I've spent the last 2 weeks refactoring one of my side projects and only bothered testing right at the end. There was only a handful of simple bugs that TypeScript and Mypy couldn't detect, and the in-browser errors told me exactly where to look to fix them.