My top ones: - abstract interpretation. It was at least a year after I first heard of the concept that I grokked it enough to try to write my own, and it wasn’

My top ones: - abstract interpretation. It was at least a year after I first heard of the concept that I grokked it enough to try to write my own, and it wasn’t until many goofy failed attempts and something like 6–7 years that I actually understood it enough to write a good one. And then it was another 5-6 years before I understood the theory well enough to understand abstract interpreters designed by others. So, like, that took me more than a decade to understand. Maybe I still don’t fully understand it. - SSA IR design. It’s so easy to understand the facts of what SSA is, and it’s not that hard to use an SSA IR designed by others. But it took me over ten years from when I first read the papers and did my first attempt until when I actually got it. And I still feel like there are aspects of SSA that I don’t fully understand.