The really interesting thing is that while I'm completely happy to spend 30 mins writing line after line of code to implement some algorithm, winding up with >1

The really interesting thing is that while I'm completely happy to spend 30 mins writing line after line of code to implement some algorithm, winding up with >100 lines and only compiling at the end, if it takes more than around 30 seconds to compile I'll lose my train of thought entirely. So I understand the concept from the point of view of task-switching. But a lot of people in the comments allude to needing a feedback loop for each and every line of code, which sounds absolutely horrifying to me - is that really what development is like in some environments? you can't really be sure of what a single line of code is going to do?