Create continuous "chunked" grids/volumes with ndarrays


An experiment with creating continuous grids/volumes from "chunked" ndarrays. Ideally, this could make it easy to work with infinite terrain, or simply splitting a large area into smaller chunks to conserve memory - while still taking advantage of the modules designed to work with ndarrays.

The interface supports both asynchronous and synchronous getters, in case you want to back the chunks with a local storage mechanism such as level.js.

Takes an options object, with the following properties:

  • shape: the shape (dimensions) of each chunk.
  • getter: a function that is called when a new is required.

The getter function takes two arguments, position and done. The former is the chunk's position (in chunks, not units), and the latter is a Node-style callback for returning the new array. If you want to retrieve chunks synchronously you can return the array as well. All chunks must be the same shape as specified in the shape option.

Retrieves a single chunk, by its position in chunkspace.

Retrieves a proxy array combining the chunks between hi and lo inclusive.

The local chunk cache - an object containing each chunk, indexed by their positions.

Clears a chunk from the local object cache - you should be doing this when you're no longer using a chunk to conserve memory.

var field = require('ndarray-continuous')({
    shape: [50, 50, 50]
  , getterfunction(positiondone) {
    return done(null, zeros([50, 50]))
var array = field.group([-5, -5, -5], [5, 5, 5])
// things happen... 
setTimeout(function() {
    .filter(function() { return true })
    .forEach(function(key) {
      var chunk = field.index[key]
}, 1000)