I usually leave "visuals" as some sort of afterthought, which is weird as I was mainly a frontend developer for years.
The last two days were dedicated to fix the rendering pipeline in the game I'm writing, from obvious AI stitched code from thing to thing just to make it work
To a full parallelizable emitter architecture with command arrays, these commands then are given to MonoGame's to do the actual drawing.
I now feel confident that any changes I make to my rendering pipeline are predictable and won't break in weird stuff
adding a new rendering feature would likely be another command array with the same parallelizable potential.
As usual, F# gives me the confidence that refactors will still leave my codebase working whenever I have good domain types driving all the thing down to the last consumer
PR for reference:
https://github.com/AngelMunoz/Kipo/pull/16