Does this also fix #3594? I made a patch to potentially fix that issue, which affects the same lines of code, but I was unsure about it's correctness in all situations.

I think it should

I looked at your patch, and the diff between yours and mine is that mine makes it so that moving light-emitting blocks don't get AO, while in yours they could get AO, but only if the quad explicitly sets AO to TriState.TRUE.

my fix could be modified further if you removed the "&& this.defaultAo" inside the else block of the processQuad method of the NonTerrainBlockRenderContext class to mimic the functionality of your patch if you wanted it to

Would that improve the situation? I'm not sure what this code does or what defaultAo is for.

My code makes it so the tesselateBlock method actually uses the global AO provided by the constructor. this.defaultAo is true when Ambient Occlusion is enabled globally in the context and if the block doesn't emit light. Removing this.defaultAo would allow light-emitting blocks that are rendered through the NonTerrainBlockRenderContext to get AO if TriState.TRUE is passed through the method. If you keep it, they will never get AO even if TriState.TRUE is passed through the processQuad method.

Removing this.defaultAo would allow light-emitting blocks that are rendered through the NonTerrainBlockRenderContext to get AO if TriState.TRUE is passed through the method.

So doing this is necessary for correctness?

It depends on if the intended functionality for TriState.TRUE is to force AO no matter whether a block is emitting light or not. If it is, then i could remove it.

I don't know what the intended functionality of this code is. I'll try to find out from other people.

Well it turns out nobody knows what this code is supposed to be doing. So I'll just merge this and if it breaks we can look into it again. Thanks for taking the time to work through this.