phantom

============== Asynchronous Scala DSL for Cassandra

Using phantom

The current version is: val phantomVersion = 1.0.0. Phantom is published to Maven Central and it's actively and avidly developed.

Issues and questions
Adopters
Roadmap
Commercial support
Using phantom in your project
phantom columns
Data modeling with phantom
Querying with phantom
Basic query examples
Collection operators
Automated schema generation
Cassandra indexing
Asynchronous iterators
Batch statements
Thrift integration
Running the tests locally
Contributing to phantom
- Using GitFlow as a branching model
- Scala style guidelines for contributions
- Using the testing utilities to write tests
Copyright

Issues and questions

We love Cassandra to bits and use it in every bit our stack. phantom makes it super trivial for Scala users to embrace Cassandra.

Cassandra is highly scalable and it's by far the most powerful database technology available, open source or otherwise.

Phantom is built on top of the Datastax Java Driver, which does most of the heavy lifting.

If you're completely new to Cassandra, a much better place to start is the Datastax Introduction to Cassandra

We are very happy to help implement missing features in phantom, answer questions about phantom, and occasionally help you out with Cassandra questions, although do note we're a bit short staffed!

You can get in touch via the newzly-phantom Google Group or via the below listed emails.

We are also extremely grateful if you add your company to our list of adopters, as it makes it easy for us to further increase adoption, contributions and make phantom even more awesome.

Adopters

This is a list of companies that have embraced phantom as part of their technology stack and using it in production environments.

Roadmap

While dates are not fixed, we will use this list to tell you about our plans for the future. If you have great ideas about what could benefit all phantom adopters, please get in touch. We are very happy and eager to listen.

User defined types

We are working closely around the latest features in the Datastax Java driver and Apache Cassandra 2.1 to offer a fully type safe DSL for user defined types. This feature is well in progress and you can expect to see it live roughly at the same time as the release of the Datastax 2.1 driver, planned for July 2014.

Some of the cool features include automatic schema generation, fully type safe referencing of fields and inner members of UDTs and fully type safe querying.

Zookeeper support

Since Cassandra cannot be loadbalanced effectively and Zookeeper is to date the de-facto standard for distributed synchronisation, we figured a pre-build integration based on finagle-zookeeper would be awesome.

We've even taken it one step further, writing some pretty cool tools for testing automations. With a simple trait you can run asynchronous tests against an embedded Cassandra instance and an embedded Zookeeper instance. This process is completely transparent and you don't really need to do anything. No config or starting tools is necessary, everything will start and stop automatically, configure itself automatically and run tests in parallel using async assertions, all automatic and with our tools.

This awesome feature is still in battle-testing while we iron out some last few bugs and nastiness from its API, you can expect to see it live around phantom 0.9.0 or 1.0.0, but it's coming together quite nicely and we have already upgraded a lot of our testing to use the fancy new tooling.

We are also testing it in production in a massive enterprise to make sure it's reliable with a few dozen nodes in a cluster, not just the local embedded flavour.

Spark integration

Thanks to the awesome partnership between Databricks and Datastax, Spark is getting a Cassandra facelift with an awesome integration. We won't be slow to follow up with a fully type safe Scala implementation of that integration, so you can enjoy the benefits of high power computation with Cassandra as a backup storage through the simple and hopefully awesome DSL we've gotten you used to.

You can expect to see the spark integration live in a new phantom-spark module in the 1.1.0 or 1.2.0 version, planned sometime in September 2014.

Commercial support

We, the people behind phantom run a software development house specialised in Scala, NoSQL and distributed systems. If you are after enterprise grade training or support for using phantom, Websudos is here to help!

We offer a comprehensive range of services, including but not limited to:

Remote contractors for hire
Training
In house software development
Team building
Architecture planning
Startup product development

We are big fans of open source and we will open source every project we can! To read more about our OSS efforts, click here.

Integrating phantom in your project

For most things, all you need is phantom-dsl. Read through for information on other modules.

libraryDependencies ++= Seq(
  "com.websudos"  %% "phantom-dsl"                   % phantomVersion
)

The full list of available modules is:

libraryDependencies ++= Seq(
  "com.websudos"  %% "phantom-dsl"                   % phantomVersion,
  "com.websudos"  %% "phantom-example"               % phantomVersion,
  "com.websudos"  %% "phantom-spark"                 % phantomVersion,
  "com.websudos"  %% "phantom-thrift"                % phantomVersion,
  "com.websudos"  %% "phantom-test"                  % phantomVersion
  "com.websudos"  %% "phantom-zookeeper"             % phantomVersion
)

Primitive columns

This is the list of available columns and how they map to C* data types. This also includes the newly introduced static columns in C* 2.0.6.

The type of a static column can be any of the allowed primitive Cassandra types. phantom won't let you mixin a non-primitive via implicit magic.

phantom columns	Java/Scala type	Cassandra type
BigDecimalColumn	scala.math.BigDecimal	decimal
BigIntColumn	scala.math.BigInt	varint
BooleanColumn	scala.Boolean	boolean
DateColumn	java.util.Date	timestamp
DateTimeColumn	org.joda.time.DateTime	timestamp
DoubleColumn	scala.Double	double
FloatColumn	scala.Float	float
IntColumn	scala.Int	int
InetAddressColumn	java.net.InetAddress	inet
LongColumn	scala.Long	long
StringColumn	java.lang.String	text
UUIDColumn	java.util.UUID	uuid
TimeUUIDColumn	java.util.UUID	timeuuid
CounterColumn	scala.Long	counter
CounterColumn	scala.Long	counter
StaticColumn<type>	<type>	type static

Optional primitive columns

Optional columns allow you to set a column to a null or a None. Use them when you really want something to be optional. The outcome is that instead of a T you get an Option[T] and you can match, fold, flatMap, map on a None.

The Optional part is handled at a DSL level, it's not translated to Cassandra in any way.

phantom columns	Java/Scala type	Cassandra columns
OptionalBigDecimalColumn	Option[scala.math.BigDecimal]	decimal
OptionalBigIntColumn	Option[scala.math.BigInt]	varint
OptionalBooleanColumn	Option[scala.Boolean]	boolean
OptionalDateColumn	Option[java.util.Date]	timestamp
OptionalDateTimeColumn	Option[org.joda.time.DateTime]	timestamp
OptionalDoubleColumn	Option[scala.Double]	double
OptionalFloatColumn	Option[scala.Float]	float
OptionalIntColumn	Option[scala.Int]	int
OptionalInetAddressColumn	Option[java.net.InetAddress]	inet
OptionalLongColumn	Option[Long]	long
OptionalStringColumn	Option[java.lang.String]	text
OptionalUUIDColumn	Option[java.util.UUID]	uuid
OptionalTimeUUID	Option[java.util.UUID]	timeuuid

Collection columns

Cassandra collections do not allow custom data types. Storing JSON as a string is possible, but it's still a text column as far as Cassandra is concerned. The type in the below example is always a default C* type.

JSON columns require you to define a toJson and fromJson method, telling phantom how to go from a String to the type you need. It makes no assumptions as to what library you are using, although we have tested with lift-json and play-json.

Examples on how to use JSON columns can be found in JsonColumnTest.scala

phantom columns	Cassandra columns
ListColumn.<type>	list<type>
SetColumn.<type>	set<type>
MapColumn.<type, type>	map<type, type>
JsonColumn.<type>	text
JsonListColumn.<type>	list<text>
JsonSetColumn.<type>	set<type>

Indexing columns

phantom uses a specific set of traits to enforce more advanced Cassandra limitations and schema rules at compile time. Instead of waiting for Cassandra to tell you you've done bad things, phantom won't let you compile them, saving you a lot of time.

PartitionKey[T]

This is the default partitioning key of the table, telling Cassandra how to divide data into partitions and store them accordingly. You must define at least one partition key for a table. Phantom will gently remind you of this with a fatal error.

If you use a single partition key, the PartitionKey will always be the first PrimaryKey in the schema.

It looks like this in CQL: PRIMARY_KEY(your_partition_key, primary_key_1, primary_key_2).

Using more than one PartitionKey[T] in your schema definition will output a Composite Key in Cassandra. PRIMARY_KEY((your_partition_key_1, your_partition_key2), primary_key_1, primary_key_2).

PrimaryKey[T]

As it's name says, using this will mark a column as PrimaryKey. Using multiple values will result in a Compound Value. The first PrimaryKey is used to partition data. phantom will force you to always define a PartitionKey so you don't forget about how your data is partitioned. We also use this DSL restriction because we hope to do more clever things with it in the future.

A compound key in C* looks like this: PRIMARY_KEY(primary_key, primary_key_1, primary_key_2).

Before you add too many of these, remember they all have to go into a where clause. You can only query with a full primary key, even if it's compound. phantom can't yet give you a compile time error for this, but Cassandra will give you a runtime one.

Index

This is a SecondaryIndex in Cassandra. It can help you enable querying really fast, but it's not exactly high performance. It's generally best to avoid it, we implemented it to show off what good guys we are.

When you mix in Index[T] on a column, phantom will let you use it in a where clause. However, don't forget to allowFiltering for such queries, otherwise C* will give you an error.

ClusteringOrder

This can be used with either java.util.Date or org.joda.time.DateTime. It tells Cassandra to store records in a certain order based on this field.

An example might be: object timestamp extends DateTimeColumn(this) with ClusteringOrder[DateTime] with Ascending To fully define a clustering column, you MUST also mixin either Ascending or Descending to indicate the sorting order.

Thrift Columns

These columns are especially useful if you are building Thrift services. They are deeply integrated with Twitter Scrooge and relevant to the Twitter ecosystem(Finagle, Zipkin, Storm etc) They are available via the phantom-thrift module and you need to import com.websudos.phantom.thrift.Implicits._ to get them.

In the below scenario, the C* type is always text and the type you need to pass to the column is a Thrift struct, specifically com.twitter.scrooge.ThriftStruct. phantom will use a CompactThriftSerializer, store the record as a binary string and then reparse it on fetch.

Thrift serialization and de-serialization is extremely fast, so you don't need to worry about speed or performance overhead. You generally use these to store collections(small number of items), not big things.

phantom columns	Cassandra columns
ThriftColumn.<type>	text
ThriftListColumn.<type>	list<text>
ThriftSetColumn.<type>	set<text>
ThriftMapColumn.<type, type>	map<text, text>

Data modeling with phantom

import java.util.{ Date, UUID }
import com.datastax.driver.core.Row
import com.websudos.phantom.sample.ExampleModel
import com.websudos.phantom.Implicits._

case class ExampleModel (
  id: Int,
  name: String,
  props: Map[String, String],
  timestamp: Int,
  test: Option[Int]
)

sealed class ExampleRecord extends CassandraTable[ExampleRecord, ExampleModel] {

  object id extends UUIDColumn(this) with PartitionKey[UUID]
  object timestamp extends DateTimeColumn(this) with ClusteringOrder with Ascending
  object name extends StringColumn(this)
  object props extends MapColumn[ExampleRecord, ExampleModel, String, String](this)
  object test extends OptionalIntColumn(this)

  override def fromRow(row: Row): ExampleModel = {
    ExampleModel(id(row), name(row), props(row), timestamp(row), test(row));
  }
}

Querying with Phantom

The query syntax is inspired by the Foursquare Rogue library and aims to replicate CQL 3 as much as possible.

Phantom works with both Scala Futures and Twitter Futures as first class citizens.

"Select" queries

Method name	Description
`where`	The `WHERE` clause in CQL
`and`	Chains several clauses, creating a `WHERE ... AND` query
`orderBy`	Adds an `ORDER_BY column_name` to the query
`allowFiltering`	Allows Cassandra to filter records in memory. This is an expensive operation.
`useConsistencyLevel`	Sets the consistency level to use.
`setFetchSize`	Sets the maximum number of records to retrieve. Default is 10000
`limit`	Sets the exact number of records to retrieve.

Select queries are very straightforward and enforce most limitations at compile time.

`where` and `and` clause operators

Operator name	Description
eqs	The "equals" operator. Will match if the objects are equal
in	The "in" operator. Will match if the object is found the list of arguments
gt	The "greater than" operator. Will match a the record is greater than the argument and exists
gte	The "greater than or equals" operator. Will match a the record is greater than the argument and exists
lt	The "lower than" operator. Will match a the record that is less than the argument and exists
lte	The "lower than or equals" operator. Will match a the record that is less than the argument and exists

Partial selects

All partial select queries will return Tuples and are therefore limited to 22 fields. We haven't yet bothered to add more than 10 fields in the select, but you can always do a Pull Request. The file you are looking for is here. The 22 field limitation will change in Scala 2.11 and phantom will be updated once cross version compilation is enabled.

  def getNameById(id: UUID): Future[Option[String]] = {
    ExampleRecord.select(_.name).where(_.id eqs someId).one()
  }

  def getNameAndPropsById(id: UUID): Future[Option(String, Map[String, String])] {
    ExampleRecord.select(_.name, _.props).where(_.id eqs someId).one()
  }

"Insert" queries

Method name	Description
`value`	A type safe Insert query builder. Throws an error for `null` values.
`valueOrNull`	This will accept a `null` without throwing an error.
`useConsistencyLevel`	Sets the consistency level to use.
`ttl`	Sets the "Time-To-Live" for the record.

"Update" queries

Method name	Description
`where`	The `WHERE` clause in CQL
`and`	Chains several clauses, creating a `WHERE ... AND` query
`modify`	The actual update query builder
`useConsistencyLevel`	Sets the consistency level to use.
`onflyIf`	Addition update condition. Used on non-primary columns

"Delete" queries

Method name	Description
`where`	The `WHERE` clause in CQL
`useConsistencyLevel`	Sets the consistency level to use.

Common query methods

The full list can be found in [CQLQuery.scala](https://github.com/websudos/phantom/blob/develop/phantom-dsl/src/main/scala/com/websudos/phantom/query/CQLQuery .scala).

Method name	Description
`tracing_=`	The Cassandra utility method. Enables or disables tracing.
`queryString`	Get the output CQL 3 query of a phantom query.
`consistencyLevel`	Retrieves the consistency level in use.
`consistencyLevel_=`	Sets the consistency level to use.
`retryPolicy`	Retrieves the RetryPolicy in use.
`retryPolicy_=`	Sets the RetryPolicy to use.
`serialConsistencyLevel`	Retrieves the serial consistency level in use.
`serialConsistencyLevel_=`	Sets the serial consistency level to use.
`forceNoValues_=`	Sets the serial consistency level to use.
`routingKey`	Retrieves the Routing Key as a ByteBuffer.

Scala Futures

ExampleRecord.select.one() // When you only want to select one record
ExampleRecord.update.where(_.name eqs name).modify(_.name setTo "someOtherName").future() // When you don't care about the return type.
ExampleRecord.select.fetchEnumerator // when you need an Enumerator
ExampleRecord.select.fetch // When you want to fetch a Seq[Record]

Examples with Scala Futures

import scala.concurrent.ExecutionContext.Implicits.global
import scala.concurrent.Future

object ExampleRecord extends ExampleRecord {
  override val tableName = "examplerecord"

  // now define a session, a normal Datastax cluster connection
  implicit val session = SomeCassandraClient.session;

  def getRecordsByName(name: String): Future[Seq[ExampleModel]] = {
    ExampleRecord.select.where(_.name eqs name).fetch
  }

  def getOneRecordByName(name: String, someId: UUID): Future[Option[ExampleModel]] = {
    ExampleRecord.select.where(_.name eqs name).and(_.id eqs someId).one()
  }
}

Twitter Futures

ExampleRecord.select.get() // When you only want to select one record
ExampleRecord.update.where(_.name eqs name).modify(_.name setTo "someOtherName").execute() // When you don't care about the return type.
ExampleRecord.select.enumerate // when you need an Enumerator
ExampleRecord.select.collect // When you want to fetch a Seq[Record]

More examples with Twitter Futures

import com.twitter.util.Future

object ExampleRecord extends ExampleRecord {
  override val tableName = "examplerecord"

  // now define a session, a normal Datastax cluster connection
  implicit val session = SomeCassandraClient.session;

  def getRecordsByName(name: String): Future[Seq[ExampleModel]] = {
    ExampleRecord.select.where(_.name eqs name).collect
  }

  def getOneRecordByName(name: String, someId: UUID): Future[Option[ExampleModel]] = {
    ExampleRecord.select.where(_.name eqs name).and(_.id eqs someId).get()
  }
}

Collections and operators

Based on the above list of columns, phantom supports CQL 3 modify operations for CQL 3 collections: list, set, map. All operators will be available in an update query, specifically:

ExampleRecord.update.where(_.id eqs someId).modify(_.someList $OPERATOR $args).future().

List operators

Examples in ListOperatorsTest.scala.

Name	Description
`prepend`	Adds an item to the head of the list
`prependAll`	Adds multiple items to the head of the list
`append`	Adds an item to the tail of the list
`appendAll`	Adds multiple items to the tail of the list
`discard`	Removes the given item from the list.
`discardAll`	Removes all given items from the list.
`setIdIx`	Updates a specific index in the list

Set operators

Sets have a better performance than lists, as the Cassandra documentation suggests. Examples in SetOperationsTest.scala.

Name	Description
`add`	Adds an item to the tail of the set
`addAll`	Adds multiple items to the tail of the set
`remove`	Removes the given item from the set.
`removeAll`	Removes all given items from the set.

Map operators

Both the key and value types of a Map must be Cassandra primitives. Examples in MapOperationsTest.scala:

Name	Description
`put`	Adds an (key -> value) pair to the map
`putAll`	Adds multiple (key -> value) pairs to the map

Automated schema generation

Replication strategies and more advanced features are not yet available in phantom, but CQL 3 Table schemas are automatically generated from the Scala code. To create a schema in Cassandra from a table definition:

import scala.concurrent.Await
import scala.concurrent.duration._

Await.result(ExampleRecord.create().future(), 5000 millis)

Of course, you don't have to block unless you want to.

Partition tokens

import scala.concurrent.Await
import scala.concurrent.duration._
import com.websudos.phantom.Implicits._

sealed class ExampleRecord2 extends CassandraTable[ExampleRecord2, ExampleModel] with LongOrderKey[ExampleRecod2, ExampleRecord] {

  object id extends UUIDColumn(this) with PartitionKey[UUID]
  object timestamp extends DateTimeColumn(this)
  object name extends StringColumn(this)
  object props extends MapColumn[ExampleRecord2, ExampleRecord, String, String](this)
  object test extends OptionalIntColumn(this)

  override def fromRow(row: Row): ExampleModel = {
    ExampleModel(id(row), name(row), props(row), timestamp(row), test(row));
  }
}


val orderedResult = Await.result(Articles.select.where(_.id gtToken one.get.id ).fetch, 5000 millis)

PartitionToken operators

Operator name	Description
eqsToken	The "equals" operator. Will match if the objects are equal
gtToken	The "greater than" operator. Will match a the record is greater than the argument
gteToken	The "greater than or equals" operator. Will match a the record is greater than the argument
ltToken	The "lower than" operator. Will match a the record that is less than the argument and exists
lteToken	The "lower than or equals" operator. Will match a the record that is less than the argument

For more details on how to use Cassandra partition tokens, see SkipRecordsByToken.scala

Cassandra Time Series

phantom supports Cassandra Time Series. To use them, simply mixin com.websudos.phantom.keys.ClusteringOrder and either Ascending or Descending.

Restrictions are enforced at compile time.

import com.websudos.phantom.Implicits._

sealed class ExampleRecord3 extends CassandraTable[ExampleRecord3, ExampleModel] with LongOrderKey[ExampleRecod3, ExampleRecord] {

  object id extends UUIDColumn(this) with PartitionKey[UUID]
  object timestamp extends DateTimeColumn(this) with ClusteringOrder with Ascending
  object name extends StringColumn(this)
  object props extends MapColumn[ExampleRecord2, ExampleRecord, String, String](this)
  object test extends OptionalIntColumn(this)

  override def fromRow(row: Row): ExampleModel = {
    ExampleModel(id(row), name(row), props(row), timestamp(row), test(row));
  }
}

Automatic schema generation can do all the setup for you.

Compound keys

Phantom also supports using Compound keys out of the box. The schema can once again by auto-generated.

A table can have only one PartitionKey but several PrimaryKey definitions. Phantom will use these keys to build a compound value. Example scenario, with the compound key: (id, timestamp, name)

import org.joda.time.DateTime
import com.websudos.phantom.Implicits._

sealed class ExampleRecord3 extends CassandraTable[ExampleRecord3, ExampleModel] with LongOrderKey[ExampleRecod3, ExampleRecord] {

  object id extends UUIDColumn(this) with PartitionKey[UUID]
  object timestamp extends DateTimeColumn(this) with PrimaryKey[DateTime]
  object name extends StringColumn(this) with PrimaryKey[String]
  object props extends MapColumn[ExampleRecord2, ExampleRecord, String, String](this)
  object test extends OptionalIntColumn(this)

  override def fromRow(row: Row): ExampleModel = {
    ExampleModel(id(row), name(row), props(row), timestamp(row), test(row));
  }
}

CQL 3 Secondary Keys

When you want to use a column in a where clause, you need an index on it. Cassandra data modeling is out of the scope of this writing, but com.websudos.phantom offers com.websudos.phantom.keys.Index to enable querying.

The CQL 3 schema for secondary indexes can also be auto-generated with ExampleRecord4.create().

SELECT is the only query you can perform with an Index column. This is a Cassandra limitation. The relevant tests are found here.

import java.util.UUID
import org.joda.time.DateTime
import com.websudos.phantom.Implicits._

sealed class ExampleRecord4 extends CassandraTable[ExampleRecord4, ExampleModel] with LongOrderKey[ExampleRecod4, ExampleRecord] {

  object id extends UUIDColumn(this) with PartitionKey[UUID]
  object timestamp extends DateTimeColumn(this) with Index[DateTime]
  object name extends StringColumn(this) with Index[String]
  object props extends MapColumn[ExampleRecord2, ExampleRecord, String, String](this)
  object test extends OptionalIntColumn(this)

  override def fromRow(row: Row): ExampleModel = {
    ExampleModel(id(row), name(row), props(row), timestamp(row), test(row));
  }
}

Asynchronous iterators for large record sets

Phantom comes packed with CQL rows asynchronous lazy iterators to help you deal with billions of records. phantom iterators are based on Play iterators with very lightweight integration.

The functionality is identical with respect to asynchronous, lazy behaviour and available methods. For more on this, see this Play tutorial

Usage is trivial. If you want to use slice, take or drop with iterators, the partitioner needs to be ordered.

import scala.concurrent.Await
import scala.concurrent.duration._
import org.joda.time.DateTime
import com.websudos.phantom.Implicits._


sealed class ExampleRecord3 extends CassandraTable[ExampleRecord3, ExampleModel] with LongOrderKey[ExampleRecord3, ExampleRecord] {

  object id extends UUIDColumn(this) with PartitionKey[UUID]
  object timestamp extends DateTimeColumn(this) with PrimaryKey[DateTime]
  object name extends StringColumn(this) with PrimaryKey[String]
  object props extends MapColumn[ExampleRecord2, ExampleRecord, String, String](this)
  object test extends OptionalIntColumn(this)

  override def fromRow(row: Row): ExampleModel = {
    ExampleModel(id(row), name(row), props(row), timestamp(row), test(row));
  }
}

object ExampleRecord3 extends ExampleRecord3 {
  def getRecords(start: Int, limit: Int): Future[Set[ExampleModel]] = {
    select.fetchEnumerator.slice(start, limit).collect
  }
}

Batch statements

phantom also brrings in support for batch statements. To use them, see [IterateeBigTest.scala]( https://github .com/websudosuk/phantom/blob/develop/phantom-test/src/test/scala/com/websudos/phantom/iteratee/IterateeBigTest.scala)

We have tested with 10,000 statements per batch, and 1000 batches processed simulatenously. Before you run the test, beware that it takes ~40 minutes.

Batches use lazy iterators and daisy chain them to offer thread safe behaviour. They are not memory intensive and you can expect consistent processing speed even with 1 000 000 statements per batch.

Batches are immutable and adding a new record will result in a new Batch, just like most things Scala, so be careful to chain the calls.

phantom also supports COUNTER batch updates and UNLOGGED batch updates.

LOGGED batch statements

import com.websudos.phantom.Implicits._

BatchStatement()
    .add(ExampleRecord.update.where(_.id eqs someId).modify(_.name setTo "blabla"))
    .add(ExampleRecord.update.where(_.id eqs someOtherId).modify(_.name setTo "blabla2"))
    .future()

COUNTER batch statements

import com.websudos.phantom.Implicits._

CounterBatchStatement()
    .add(ExampleRecord.update.where(_.id eqs someId).modify(_.someCounter increment 500L))
    .add(ExampleRecord.update.where(_.id eqs someOtherId).modify(_.someCounter decrement 300L))
    .future()

UNLOGGED batch statements

import com.websudos.phantom.Implicits._

UnloggedBatchStatement()
    .add(ExampleRecord.update.where(_.id eqs someId).modify(_.name setTo "blabla"))
    .add(ExampleRecord.update.where(_.id eqs someOtherId).modify(_.name setTo "blabla2"))
    .future()

Thrift integration

We use Apache Thrift extensively for our backend services. phantom is very easy to integrate with Thrift models and uses Twitter Scrooge to compile them. Thrift integration is optional and available via "com.websudos" %% "phantom-thrift" % phantomVersion.

namespace java com.websudos.phantom.sample.ExampleModel

stuct ExampleModel {
  1: required i32 id,
  2: required string name,
  3: required Map&lt;string, string&gt; props,
  4: required i32 timestamp
  5: optional i32 test
}

Running the tests locally

phantom uses Embedded Cassandra to run tests without a local Cassandra server running. You need two terminals to run the tests, one for Embedded Cassandra and one for the actual tests.

sbt
project phantom-cassandra-unit
run

Then in a new terminal

sbt
project com.websudos.phantom-test
test

Contributors

Phantom was developed at websudos as an in-house project. All Cassandra integration at websudos goes through phantom.

Flavian Alexandru flavian@newzly.com(maintainer)
Tomasz Perek tomasz.perek@newzly.com
Viktor Taranenko (viktortnk)
Bartosz Jankiewicz (@bjankie1)
Eugene Zhulenev (@ezhulenev)

Copyright

Special thanks to Viktor Taranenko from WhiskLabs, who gave us the original idea.

Contributing to phantom

Contributions are most welcome!

Using GitFlow

To contribute, simply submit a "Pull request" via GitHub.

We use GitFlow as a branching model and SemVer for versioning.

When you submit a "Pull request" we require all changes to be squashed.
We never merge more than one commit at a time. All the n commits on your feature branch must be squashed.
We won't look at the pull request until Travis CI says the tests pass, make sure tests go well.

Scala Style Guidelines

In spirit, we follow the Twitter Scala Style Guidelines. We will reject your pull request if it doesn't meet code standards, but we'll happily give you a hand to get it right.

Some of the things that will make us seriously frown:

Blocking when you don't have to. It just makes our eyes hurt when we see useless blocking.
Testing should be thread safe and fully async, use ParallelTestExecution if you want to show off.
Writing tests should use the pre-existing tools, they bring in EmbeddedCassandra, Zookeeper and other niceties, allowing us to run multi-datacenter tests.
Use the common patterns you already see here, we've done a lot of work to make it easy.
Don't randomly import stuff. We are very big on alphabetized clean imports.

Using phantom

Adopters

Roadmap

Commercial support

Contributing to phantom

About

Languages