• added MultiPrepareBuilder for creating MultiPrepare instance for MultiMap from Java.
• renamed Bag.Less to Glass and Bag.less instance to Bag.glass. Glass being a transparent type.
• removed redundant BAG[_] type param from getKeyValueDeadline functions
• bumped ZIO version to 1.0.2

Release overview

• Improved SkipList implementation for faster writes and atomic reads
• Removed snapshotting LevelZero state for reads.

Improved write performance for sequential writes

Implemented a faster NavigableMap (for SequentialOrder writes) backed by Array and binary-search which can be enabled for applications that only require sequential writes and do not need random writes.

The following shows benchmarks of SequentialOrder (Array) vs RandomOrder (java.ConcurrentSkipListMap) for sequential writes.

How to optimise SwayDB for SequentialOrder writes?

Default setting optimises writes for RandomOrder. The following code samples optimise writes for SequentialOrder.

initialSkipListLength sets the length of the Array since SequentialOrder is backed by Array.

Java

Map<Integer, String, Void> map =
  MemoryMap
    .functionsOff(Default.intSerializer(), Default.stringSerializer())
    .setOptimiseWrites(OptimiseWrites.sequentialOrder(10000))
    .get();

Scala

val map = 
  memory.Map[Int, String, Nothing, Bag.Less](
    optimiseWrites = OptimiseWrites.SequentialOrder(initialSkipListLength = 10000)
  )

Atomic SkipList reads

Previously SkipList required copying in-memory key-values for atomicity (#124). Now LevelZero performs atomic reads without copying.

Default setting disables Atomic which can be enabled with the following.

Java

Map<Integer, String, Void> map =
  MemoryMap
    .functionsOff(Default.intSerializer(), Default.stringSerializer())
    .setAtomic(Atomic.on())
    .get();

Scala

val map =
  memory.Map[Int, String, Nothing, Bag.Less](atomic = Atomic.On)

Removed snapshotting

Previously reads were creating a snapshot which provided the state of write-ahead-log files in LevelZero for each read. Now snapshotting happen lazily (Stream like) as the read progresses. This change was aimed to further minimise GC workload.

This release was focused on

• MultiMap & Compaction performance improvements.
• Completing Java API.
• QA and Usability testing.
• Support for Boopickle serialisation.
• Resolving issues labeled Production release.

MultiMap and SetMap Java types

MultiMap and SetMap are now implemented for Java. Java API is now complete.

MultiMap allows creating deeply nested Map.

//create a root map
MultiMap<String, Integer, String, Void> root =
  MemoryMultiMap.functionsOff(
    stringSerializer(), //nested Map key type
    intSerializer(), //key type
    stringSerializer() //value type
  ).get();

//first child map
MultiMap<String, Integer, String, Void> child1 = root.child("first child");
child1.put(1, "child1's value 1");

SetMap provides Map like API for Set. Useful for application that always fetch the value with the key. It stores both key & value in the same location for faster seeks and reduced IOps.

SetMap<Integer, String> setMap =
  MemorySetMap
    .config(intSerializer(), stringSerializer())
    .get();

setMap.put(1, "one");

Improved MultiMap performance and reduced storage cost

MultiMap previously stored parent key bytes within child key bytes. Now deeply nested Map do not have any extra storage cost over a root MultiMap. This also improved read and write performance.

Deeply nested MultiMap has constant performance regardless of the depth.

Access to all nested child MultiMap nodes

//get all children of root map
Stream<MultiMap<String, Integer, String, Void>> children = root.children();
//get all children of root map and all it's grand children. 
Stream<MultiMap<String, Integer, String, Void>> child = root.childrenFlatten();

Functions

Shorter syntax for enabling functions

Java

Create functions without need to specify the return type. All java functions types are under PureFunctionJava.

//function that appends "updated" to old value
OnValue<Integer, String> myFunction =
  (String value) ->
    Apply.update(value + " updated");

Register the function as a List.

Map<Integer, String, PureFunction<Integer, String, Apply.Map<String>>> map =
  PersistentMap
    .functionsOn(
      Paths.get("my_map"), //directory
      Default.intSerializer(), //key serialiser
      Default.stringSerializer(), //value serializwr
      Arrays.asList(myFunction) //functions list
    ).get();

Scala

Syntax shortened for enabling functions - PureFunction.Map[Int, String]. All Scala function types are under PureFunctionScala.

val map = persistent.Map[Int, String, PureFunction.Map[Int, String], Bag.Less]("my_map")

To register function supply implicit functions List.

//sample function that appends "updated" to old value
val myFunction: OnValue[String] = (value: String) => Apply.Update(value + " updated")

implicit val myFunctions = Functions[PureFunction.Map[Int, String]](myFunction)

Added new function types

• OnKey - reads the key
• OnKeyDeadline (Scala) & OnKeyExpiration (Java) - reads the key and deadline
• OnKeyValue - reads the key and value
• OnValue - reads the value
• OnValueDeadline (Scala) & OnValueExpiration (Java) - reads the value and deadline
• OnKeyValueValueDeadline (Scala) & OnKeyValueValueExpiration (Java) - read the key, value and deadline.

Improve Serialisers

Java

Use Slice<T> to easily build custom serialisers. ByteOps.Java removes the need to cast java.lang.Byte to scala.Byte during runtime.

new Serializer<String>() {
  @Override
  public Slice<Byte> write(String data) {
    return Slice.writeStringUTF8(data, ByteOps.Java());
  }

  @Override
  public String read(Slice<Byte> slice) {
    return slice.readStringUTF8(ByteOps.Java());
  }
};

Scala

Easily serialise case classes using Bookpickle

import swaydb._
import boopickle.Default._

case class MyClass(id: Int, value: String)

implicit val serialiser = swaydb.serializers.BooPickle[MyClass]

val set = memory.Set[MyClass, Nothing, Bag.Less]()

Better Scala and Java interop

Scala functions can be shared with Java and vice versa. Scala instance can also be accessed from Java.

Set<Integer, Void> mySet =
  MemorySet
    .functionsOff(intSerializer())
    .get();

mySet.asScala(); //get scala set

Validating missing functions

Missing applied functions reported on re-boot to ensure safe startups.

swaydb.Exception$MissingFunctions: Missing 1 function. List("myFunction").

Flags for clearing applied functions

Functions that are applied can be cleared by setting the clearAppliedFunctionsOnBoot flag.

Scala

val map = persistent.Set[Int, PureFunction.Set[Int], Bag.Less]("my_set", clearAppliedFunctionsOnBoot = true)

Java

Set<Integer, PureFunction<Integer, Void, Apply.Set<Void>>> mySet =
  PersistentSet
    .functionsOn(Paths.get("my_set"), intSerializer(), Arrays.asList())
    .setClearAppliedFunctionsOnBoot(true)
    .get();

Build Validator - Error reporting

Incompatible SwayDB version files get reported on boot-up.

Eg: Current v0.16 is incompatible with older versions so start v0.16 on older versions will result in the following error

Incompatible versions! SwayDB v0.16 is not compatible with files created by v0.15

Directories are type-checked. Eg: starting Set on a Map directory will yield

Invalid data type Set for the directory of type Map.

• Streams created from data-types (Map, Set, Queue ... ) use the bag from within the data-type.
• added eventually.persistent.SetMap
• added added eventually.persistent.Queue
• added build.info file which stores the version and used to perform compatibility checks.
• MultiMap is not experimental anymore.

Breaking change - with the addition of build.info file, previous versions are incompatible with this version.

Resolved issues

#260 - Rename MapConfig, SetConfig & QueueConfig in Java to reflect their storage types #259 - Remove Function type param from SetMap #258 - Java's Set.stream is returning Stream instead of Source #257 - Add asJava to interop with java.util.Map #256 - Terminate Bag on close #255 - Implement Source for Stream #253 - Add Stream.materializeBuilder for custom collection type #252 - SwayDBExpireSpec test failing

This release fixes issues relating to memory-mapped files not working on Windows and other important issues.

#156 - program won't end (resolved) & java.nio.file.AccessDeniedException on Windows #251 - [WINDOWS] - MappedByteBuffer.force is extremely slow on Windows Performance #250 - Compaction does not progress on reboot - ArrayIndexOutOfBoundsException #249 - ByteBufferSweeper Actor can be a basic Actor

See issue #156 for more details on new ForceSave and MMAP configurations to handle memory-mapped files on Windows.

Resolves important issues.

• #248 - Disallow api calls after close or delete
• #247 - Releasing FileLock does not close channel
• #246 - Scheduler should be internal to the Actor
• #245 - Memory mapped files result in "Too many opened files"
• #243 - BufferCleaner should be passed explicitly
• #242 - Terminate all Actors and Cache on Shutdown
• #241 - terminateAndRecover function for Actor

#240 - java and data-java modules should not disable cross builds.

Reverts change made to java artifact id in previous release - v0.14.3.

<dependency>
    <groupId>io.swaydb</groupId>
    <artifactId>java_2.13</artifactId>
    <version>0.14.4</version>
</dependency>

Resolves following bugs.

#238 - LevelZero's compaction is not updating Maps.currentMapsSlice safely #236 - Stream returns missing results when transitioning level0 entries to level1 for databases with mapSize = 1.byte #237 - Actor pauses processing messages #239 - Convenience functions for Java to overwrite default Configs parameters from Scala.

Java's artifactId changed to java instead of java_2.13

Maven dependency looks like

<dependency>
    <groupId>io.swaydb</groupId>
    <artifactId>java</artifactId>
    <version>0.14.3</version>
</dependency>

Fixes

#237 - Actor pauses processing messages