DISCLAIMER: These notes are from the defunct k8 project which precedes SquirrelJME. The notes for SquirrelJME start on 2016/02/26! The k8 project was effectively a Java SE 8 operating system and as such all of the notes are in the context of that scope. That project is no longer my goal as SquirrelJME is the spiritual successor to it.
Appears that there are some internet searches with fillPolygon being slow after a few polygons are rendered.
I could also just use float and keep just bring in my raw triangle drawer instead.
Now I have some stuff rendered, need to fix turning and offsets since it is sort of mathematical, sort of screen, and otherwise not.
The graphical drawer is right handed. My chunk layout is reverse some other hand. Walking forward moves the cubes up, walking backwards moves the cubes down. Strafe left goes left, strafe right goes right. Falling brings cubes closer, rising moves them further away. Appears that I may be looking straight down. I need to put a marker to determine which block I am actually looking at.
Traversing to draw faces takes quite awhile when there are gigantic areas to draw and such.
Trying to fix the renderer, I have discovered the origin in good view at (-15.1, 3.1, -20.3). The axis and the small unit cube appear drawn correctly while the chunk cube is a bit distorted.
Found another view at (-18.9, 12.5, -21).
Appears with the renderer right now, I am missing parts of the rectangle which make up the coordinates of the faces. One of the points ends up being really low on the chunk edge. The Z axis is also flipped.
Appears that positioning is a bit off. When I should be in a liquid block it appears that I am in a solid one.
I could use longs for positional data everywhere. Since there are 21 bits of chunk identities for each axis that means 2,097,152 chunks in one direction. Multiply that by 16 and you have 33,554,432 blocks in each direction. This gives you 37,778,931,862,957,161,709,568 blocks to work with in the world. If I use 32-bit ints for positional data in objects, this means I can have 64 sub positions in a tile for less than zero movement. So this means a 6-bit fraction followed by a 25/26 bit integral value. This means the lowest representable unit would be 0.015625. I would need a dedicated class for this representation of position. Everything else about the world such as the offsets of vertex faces would be based off of this. It would be faster for systems which lack floating point however. So rather than just longs, I could instead go for integers since the world is not explorable outside of the bounds anyway. If I do it just right, it would be possible to walk to the other side of the world from the other side, essentially just looping around. That could work though.
I can have positions and stuff be very absolute with just a single shift set for each coordinate.
The math I calculated before is not correct, there is more sub block precision than 6 bits, I believe so anyway.
When more than one world is in place, I can have a dark world. It can take the map data from the normal world and modify it a bit. So if someone has been toying around in the nature world but finds a portal to the dark world, then they will see their crafty work sort of. It will be mostly similar except with some changes based on the world layout.