server.arena

fn capture(mut src BlockSource, box Box) !&Snapshot

capture reads every block in box from src into a fresh Snapshot. Returns an error if the box exceeds max_volume so a runaway box can't exhaust memory.

fn new_box(x1 int, y1 int, z1 int, x2 int, y2 int, z2 int) Box

new_box normalizes the two corners so the caller need not order them.

interface BlockSink {
mut:
	set_block_id(id int, x int, y int, z int)
}

BlockSink writes a block by network id and broadcasts the change to viewers. Restore goes through this so a reset looks identical to normal block edits.

interface BlockSource {
mut:
	get_block(x int, y int, z int) int
}

BlockSource reads block network ids from a world by absolute coordinates. Hub satisfies it; tests use an in-memory grid.

struct Box {
pub:
	min_x int
	min_y int
	min_z int
	max_x int
	max_y int
	max_z int
}

Box is an axis-aligned block region, normalized so min <= max on every axis.

fn (b Box) volume() i64

volume is the number of blocks the box covers (inclusive on both corners). Computed in i64 so a huge box can't overflow int32 and slip past the cap.

struct Snapshot {
pub:
	box Box
mut:
	ids []int
}

Snapshot holds the block ids of a Box captured at some point in time. It is a flat array so a full restore is a single linear pass. Memory cost is 4 bytes per block - a 64^3 arena is ~1 MB.

fn (s &Snapshot) restore(mut sink BlockSink)

restore writes every captured block back through sink in the same order it was captured, so viewers see the arena reset to its saved state.

fn (s &Snapshot) len() int

len is the number of blocks stored in the snapshot.