Every now and then I check back in on the HMH streams hoping that Casey will be moving onto game the actual game, but every time I check back it's yet another experiment, problem, improvement etc.
I get that the streams are supposed to show the unfiltered process of making a game, but I feel as though it's turned more into an unfiltered stream of Casey getting side-tracked more than anything now.
It's a shame because the game play and cross-platform parts of programming a game, in my personal opinion at least, are the parts that most people struggle with, so focusing so heavily on all of these experiments with OpenGL and changes to the way the game is rendered etc. feels like a massive waste of everyone's time.
HMH is not a good source to learn game-development from.
It's more like an entertaining show, covering some highly technical details - most of which you are never-ever going to need in your game.
Crossplatform stuff is and has been trivial to do for a long time now. Unless you explicitly prohibit yourself from using external libs like SDL, GLFW and whatnot.
There's actually a very minuscule amount of technical code that you have to write, before you can start focusing solely on game-code. Without using any engines either, just plain OpenGL, GLFW and stb_image, plus some other single header kind of utility lib. Even if the game is 3D!
There are three areas where it pays off to invest some tech-time in:
1. Code hotswapping
2. Making sure that all the art / game-assets are hotswappable
3. If you're using C/C++, some sort of meta-data reflection system are in order
Casey actually touched some of these topics, but very briefly. Not in proportion to how important they are.
Everything else is a distraction unless the game you're making are explicitly technical (very few games are like this).