I actually gave up on Godot early last year because of frustrations trying to get basic in-or-under-documented things working. Even simple things like getting 2D sorting working correctly so that sprites overlap correctly was surprisingly hard (documentation said to enable y-sort flag, but in reality there were some gotchas that took me a long time to figure out). It’s a nice engine and it’s constantly improving, but it’s not really as user friendly as people make it out to be (or at least it wasn’t in early 2018, it might be a lot better now on that front). I even contributed to GDNative so I was somewhat bought into it, but in the end it wasn’t worth it for what I wanted to do. I’ll probably check it out again in the future (for now I’m too busy with non-gamedev stuff)
I gave up two times but I wanted to learn game development and I'm planning to spend the next 10 years into it. I thought Godot is my correct choice but the docs are hard to read I agree but lot of youtube tutorials out there and at this point YouTubers covered pretty much everything to create AAA or platformer game.
It's often easier to just try random things and see what works than to find answers in the documentation. It usually does not miss topics or tutorials, but the API is incomplete.
The good part is that API documentation is the easiest one to create. The troubling part is that API documentation is the easiest one to create...
I was bitten by navmesh merging in 3D, say you have several loaded scenes, each with a navmesh. If the navmesh points are matching up to an undocumented degree, the navmeshes will be merged. This is dependent upon the scene own Transform, as floating point will yield slightly different float results depending on how the navmesh point are transformed to world space. I ended up generating navmesh from an apriori description instead of Godot scenes. "optimized" path-finding into these navmesh was surprisingly poor too (considering a lot of the rest is great).
The problem is that none of the problems are documented.
