v4.0.0

We're sticking to proper semver now, so this is a much less radical change than v2 to v3 was 😀

Breaking changes

• JSON traversal is now a lot simpler! If you were only using the built-in document table functionality, no need to worry since that's the same; but if you were building criteria objects with keys like my_json_field->>innerField, you can now use standard JavaScript . and [] notation and Massive will take care of the rest.
• find has been standardized on criteria objects, and will no longer accept a * wildcard.
• Loader filters are also standardized: schema filters no longer accept all or * wildcards.

New stuff

Only one item in here, but it's a big one: resultset decomposition has arrived! Check out the documentation for a full description, but otherwise I'll let the testcase speak for itself:

const issues = yield db.everything.find({}, {
  decompose: {
    pk: 'user_id',
    columns: {
      user_id: 'id',
      username: 'username'
    },
    tests: {
      pk: 'test_id',
      columns: {
        test_id: 'id',
        name: 'name'
      },
      array: true,
      issues: {
        pk: 'id',
        columns: {
          id: 'id',
          user_id: 'user_id',
          test_id: 'test_id',
          description: 'description'
        },
        array: true
      }
    }
  }
});

assert.deepEqual(issues, [{
  id: 1,
  username: 'alice',
  tests: [{
    id: 1,
    name: 'alice\'s test',
    issues: [{
      id: 1,
      user_id: 1,
      test_id: 1,
      description: 'alice\'s issue'
    }]
  }]
}, {
  id: 2,
  username: 'bob',
  tests: []
}, {
  id: 3,
  username: 'carol',
  tests: [{
    id: 2,
    name: 'carol\'s first test',
    issues: []
  }, {
    id: 3,
    name: 'carol\'s second test',
    issues: [{
      id: 2,
      user_id: 3,
      test_id: 3,
      description: 'carol\'s issue'
    }]
  }]
}]);