Skip to content


Subversion checkout URL

You can clone with
Download ZIP
leveldb persistence for scuttlebutts (scuttlebutt/crdt/append-only and friends)
Fetching latest commit...
Cannot retrieve the latest commit at this time.
Failed to load latest commit information.


Plugin to add persistence and querying scuttlebutt documents stored in leveldb.

Instead of representing an object as a single document, scuttlebutt represents a document as a series of immutable transactions. The 'document' is modified by appending a new transaction. Old transactions that are no longer relevant can be cleaned up, but you can never modify a transaction in place. As it turns out, leveldb (a log-structured merge tree) is optimized for exactly this sort of data.


//a scuttlebutt model.
var Model = require('scuttlebutt/model')

//level-scuttlebutt needs to have an unique identifier of the current instance
var udid = require('udid')

//create a leveldb instanc
var levelup = require("levelup")
var db = levelup(DB_FILE)

//patch it with level-scuttlebutt.
var level_scuttlebutt = require("level-scuttlebutt")
level_scuttlebutt(db, udid, function (name) {
  //create a scuttlebutt instance given a name.
  //the key will match the start of the name.
  return new Model()
  //now is a good time to customize the scuttlebutt instance.

//see below...

//open a scuttlebutt instance by name., function (err, model) {
  model.on('change:key', console.log) //...
  model.set('key', value)


Add level-scuttlebutt plugin to the db object var level_scuttlebutt = require('level-scuttlebutt'); level_scuttlebutt(db, ID, schema)

ID is a unique string that identifies the node instance and should be tied to the leveldb instance. I suggest using udid.

schema should be a function that takes a string (the name of the scuttlebutt instance) and returns and empty scuttlebutt instance. You can use scuttlebutt-schema.

Map Reduce

create a map-reduce db.scuttlebutt.addMapReduce(mapReduceOptions). This is the same as db.mapReduce.add(mapReduceOptions) in map-reduce except that you should leave off start and end, and that in map: function (key, value) {...} value will be your scuttlebutt instance.

Do not assign event listeners during the map function.


get the 10 latest documents edited.

  name: 'latest10',
  map: function (key, sb) {
    var name = sb.meta.get('name') || 'no_name'
    //emit 0-many group-value pairs.
    //value must be a string or a buffer.
    this.emit([], JSON.stringify({name: name, time:, length: sb.text.length}))
  //merge the latest value into the accumulator.
  reduce: function (acc, value) {
    var all = parse(acc).concat(parse(value))
    //sort by time, decending.
    all.sort(function (a, b) {
      return b.time - a.time
    //top ten most recent
    var all = all.slice(0, 10)
    return JSON.stringify(all)
  //the first value for the accumulator.
  //since we are parsing it, it needs to be valid JSON.
  initial: '[]'



Something went wrong with that request. Please try again.