Awesome

Riakka

Scala library for talking to Riak.

Getting started

You need to have SBT installed. Please follow setup instructions.

git clone git://github.com/frank06/riakka.git
cd riakka
sbt
> update
> test

It assumes you're running Riak in localhost at port 8098, and will use a random bucket starting with riakka to run the suite.

Riak

First off, you should be familiar with Riak. There is information available here. Some other resources of interest:

• Riak Presentation at NYC NoSQL
• Intro to Riak screencast
• Riak Demo: Stickynotes
• Module jiak_resource Edoc
• ...and we're all looking forward to a wiki :)

Quick overview

Riak (through its HTTP interface, Jiak) exposes documents with some associated metadata (all in JSON format). For example:

{"bucket":"bucket", "key":"foo", "object":{"bar":"baz"}, "links":[["bucket","key","tag"], ["b2","key2","tag2"]]}"

Riakka will consistently return this structure splitted up in two: A tuple consisting of

• a metadata object (of type %) that holds important data such as bucket, key, vclock, links.
• the actual object (of type JObject) => see lift-json

Most query methods will also require a % object. There is a convenience method for creating those, as you will see in the following examples.

Features by example

Fire a Scala console (within sbt):

    > console

• first off, all the ceremony (wish it was more succint) scala> import riakka._ scala> import Jiak._ scala> import net.liftweb.json._ scala> import JsonParser._ scala> import JsonAST._ scala> import JsonDSL._

• let's initialize the client and save an object

   scala> val db = Jiak.init
   scala> val json = parse(""" {"my":"json", "list": [1, 2, 3]} """)
   scala> val metadata = %('test -> "mykey") // here we prepare a metadata object, with bucket test and key mykey
   scala> db save (metadata, json)
  

• create a link (note the save_with_response method, that returns updated metadata!)

   scala> val linked_json = ("am_i_being_linked?" -> true) ~ ("a", "b")  // a neat DSL to create JSON
   scala> val (linked_json_metadata, _) = db save_with_response (%('test -> "linked"), linked_json)  // we persist that one
   scala> val link = Link(linked_json_metadata)  // we create the link _from_ a metadata object
   scala> db save (metadata.link_+(link), json) // we pass-in a new metadata object, which contains the link
  

• walk the original object and retrieve the linked one

   scala> val linked_objects = db walk (metadata, ^^('test)) // check the docs for ^^ / link-walking spec
   scala> val (_, first_linked_object) = linked_objects first
   scala> first_linked_object.to_json // right?
  

• now clean up

   scala> db delete metadata // will use the metadata to locate the document and delete it
   scala> db delete linked_json_metadata
   scala> db get metadata // does the result make sense?
  

There is also support for If-None-Match, attachments and more. Have a look at RiakkaSpec.scala.

To be done

• Much better support of the (known) HTTP API, including schemas (read/write masks), allow_mult, specifying R/W values and datastore info.
• Support for entities (as well as plain structs made up of Lists, Tuples, etc => all things convertable to JObject), transparent serialization/deserialization to/from JSON. Should be not a big deal, as lift-json supports it wonderfully.
• Make concepts more straightforward, improve documentation and test coverage, as usual.

Needless to say, fork and send pull requests. And make use of Github's issue tracker, too.

Citing Paul from Ruby's RiakRest: "Go forth and Riak!"