kompack

0.0.2-t2indexed

Serialization engine optimizes size and performance using MessagePack binary format. Features low-level API, stream handling, and performance comparison with JSON, focusing on speed and memory efficiency.

JVMNative·dedee/kompack

Stars

—

Used by

dependents

—

Health

/ 100

Kompack - Kotlin Multiplatform MessagePack Serialization Engine

Serialization engine for Kotlin Multiplatform (JVM/native) using MessagePack binary format to optimize size and performance.

If you want to learn more about MessagePack, check this website: https://www.msgpack.org

NOTE This library is development - use with care. I am looking forward to your feedback. Current releases just contain JVM and mingw_x64 native libraries. More platforms will be built soon.

Kotlinx Serialization API

Here is an example. Simple address book structure, using Kotlin data classes with @Serializable annotation.

https://github.com/dedee/kompack-sample-application

Add dependency

dependencies {
  implementation group: 'com.wunderbee.kompack', name: 'kompack-kotlin-serialization', version: '0.0.2'
}

Use/Create your Kotlin data classes with kotlinx.serialization

@Serializable
data class Entry(val name: String, val street: String, val zip: Int, val city: String)

@Serializable
data class AddressBook(val : List<Entry>)

val addressBook = (
    (
        (, , , ),
        (, , , ),
    )
)

Serialize them

In memory (byte array) you can simply serialize the address book into a ByteArray.

val encodedByteArray = MessagePackEncoder.encodeToByteArray(addressBook)

You can also write it into a stream (e.g. file stream), which is perfect for large structures.

BufferedOutputStream(FileOutputStream("test.dat")).use { out ->
    MessagePackEncoder.encodeToStream(addressBook, out)
}

Comparison to JSON

Using MessagePack may make sense to you, if size and encoding speed matters In 'real world' scenarios, you can simply reduce the encoded size.

    val encodedByteArray = MessagePackEncoder.encodeToByteArray(addressBook)
    println("Encoded address book into ${encodedByteArray.size} bytes")
    // Encoded address book into 89 bytes

    val jsonString = Json.encodeToString(addressBook)
    println("Encoded address book into ${jsonString.length} chars")
    // Encoded address book into 179 chars

For sure, its binary and no more readable/editable. So it depends on your use case.

Low level API

In kompack-base we provide a low level API to read and write MessagePack structures. You can use that, if you do not like to use the kotlinx serialization framework.

You can serialize many different types directly into a byte array (in memory)

val byteArray = InMemoryPacker(100)
  .pack(true)
  .pack(10)
  .build()

val p = InMemoryUnpacker(byteArray)
val b = p.unpackBoolean() // true
val i = p.unpackInt() // 10

Or you can use the stream API to write larger files

StreamUnpacker(BufferedInputStream(FileInputStream("in.dat"))).use { p ->
  val b = p.unpackBoolean() // true
  val i = p.unpackInt() // 10
}

Goals

Provide Kotlin implementation of http://msgpack.org specification
Provide kotlinx.serialization engine using MessagePack format
Focus on performance and reduce memory consumption
Support Kotlin multiplatform and Native (beside JVM)
Minimal dependencies to other libraries

Performance comparison

Encoding speed using bytearray (in memory)

com.wunderbee.kompack.testsuite.comparision.Comparision

Sonar

https://sonarcloud.io/project/overview?id=dedee_kompack

Open

Support more target platforms. Build, cross compilation TBD
Array/Map type improvements Kotlin generics / reified?
Exception concept in base
BigNums not yet implemented - uint64... long
Gradle build improvements
CDB impl

Related libraries

Surfaced from shared tags and platforms — no rankings paid for.

moko-network★ 154

icerockdevFacilitates network operations by generating API classes and entities from OpenAPI specifications, supports token-based authentication, error handling, and token refreshment, enhancing networking capabilities.Shared: serialization, kotlin-native

kotlin-fhir★ 63

ohs-foundationLean, fast HL7 FHIR data-model implementation with minimal generated classes, JSON-only serialization, multi-version support, and a tiny runtime footprint for efficient healthcare data handling.Shared: serialization, kotlin-native

mp_stools★ 55

sergeychPortable utilities: sprintf-style formatting with advanced flags, base64 (including URL-safe), fast Boyer–Moore byte search, ByteArray int ops, cached expressions, reentrant coroutine mutex, coroutine-based async logging.Shared: serialization, kotlin-native

elasticmagic-kt★ 46

anti-socialAsynchronous Elasticsearch query DSL enabling typed query expressions, sub-field navigation, and declarative query modifications. Supports multiple serialization libraries and autodetects Elasticsearch and Opensearch versions.Shared: serialization, kotlin-native

snakeyaml-engine-kmp★ 42

krzema12Processes YAML 1.2 with fine-grained control and advanced functionality. Suitable for projects needing customization and flexibility. Integrates with various platforms and uses comprehensive YAML test suites for validation.Shared: serialization, kotlin-native

kotlinx-serialization-csv★ 38

hfhbdSerialize and deserialize ordered CSV and Fixed Length Format Files, leveraging a specific serialization library. Supports multiple platforms and offers easy integration via MavenCentral.Shared: serialization, kotlin-native