The key difference is that my project was a video game (this one: https://www.youtube.com/watch?v=YwXl8lDrxn8) and that takes a lot of temptation to do what the OP is writing about, rewriting stuff in Haskell for example. The interesting bit when it comes to side projects (and now my full time projects) for me isn't actually the tools or the methodology or anything like that, but the end result. I spent my teenage years building games that I never finished, and it made me quite miserable, because I so just wanted to finish something. Anything.
I remember the first game I ever finished was a clone of Snake. There was nothing remotely interesting in the implementation or in the design, but I was so flipping pleased with myself, I had finished something! And that let me slowly build my projects up to more interesting endeavours, and eventually to one that could be sold. Starting small was the key for me, as I learned two things: 1 - what was 'too much' for me to try and tackle at the time and 2 - self-discipline, not chucking the project the moment I got bored, or had a 'better idea'.