Тsфdiиg

I just tried Standard ML and it's straight up a better OCaml. I have no idea why barely anybody talks about it.

Bubblesort is famously, or infamously, the "never use this" algorithm. Donald Knuth even said it has "nothing to recommend it, except a catchy name." But there is actually a niche where it does well - game development (source linked). What makes bubblesort interesting - each step is fast and easily suspendable, and every swap leaves the array more ordered than before. Other sorts can temporarily move values away from their final positions. This matters when we render objects in a game. We want to render objects closest to the camera first (so we can skip rendering hidden objects), but we also cannot spend too much time sorting because we have strict real-time constraints. With bubblesort, we run a little bit each frame. The array gets progressively more ordered without ever blocking our game loop. It does not need to be perfectly sorted - just "better than before" is good enough. This also comes in handy while animating particles sorting into a rainbow (refer to the source for details). Each bubble sort pass becomes a single animation frame, and particles smoothly drift into place. Pretty interesting. We probably never need it, but it is always a nice reminder that "universal" rules always have their exceptions. If you are interested in digging deeper, I have linked the source article below. Give it a read.

Some time ago I coded a TUI Minesweeper in Pascal. Today I coded an AI agent in C to play that Minesweeper for me so I have more time to increase the shareholder's value. This is only the beginning. We are not ready for what's coming.