Pbandk is a Kotlin Multiplatform code generator and runtime for Protocol Buffers.
NOTE: This is the documentation for the version of pbandk currently in development. Documentation for the latest stable version is available at https://github.com/streem/pbandk/blob/v0.16.0/README.md.
- Generate idiomatic Kotlin code from protobuf definitions
- Support all platforms that are supported by Kotlin Multiplatform (JVM, Android, iOS, JS, WasmJs, WasmWasi, Linux, Windows, etc.)
- Provide an API (both pbandk's public API and generated code) that is easy to use from other Kotlin Multiplatform code and follows Kotlin conventions
- NOTE: We make a best effort to provide an API that is also easy to use directly from platform-specific code (Swift, Javascript, Typescript, Java). However, Kotlin Multiplatform interoperability with other languages still has many limitations that make this difficult to implement in practice. Pbandk's main focus is on supporting Kotlin developers. But issues and pull requests aimed at improving the pbandk API exposed to other languages are still welcome. Also see below for platform-specific notes.
- Full protobuf conformance test compatibility
- Binary and source backwards compatibility for using pbandk and pbandk-generated code from Kotlin
- Pbandk libraries follow standard semantic versioning conventions (code compiled against an older version of pbandk should be compatible with newer pbandk versions at runtime)
- Code generated by an older version of pbandk should be compatible with newer pbandk versions at runtime
- Backwards compatible changes to the proto definitions should result in binary and source compatible changes in the generated code
- Until version 1.0.0, the focus is on protobuf feature parity and correctness. Performance, code size, memory impact, etc. are of course always considered as well, and contributions in these areas are welcome. But they will not be a primary focus until after 1.0.0.
- Clean data class generation
- Works on JVM, Android, and Native (iOS, macOS, Linux, Windows, etc.), with experimental support for JS and Wasm
- Support for proto2 and proto3 syntaxes (support for Protobuf Editions is planned)
- JSON serialization/deserialization following the proto3 JSON spec (see #72 for some corner cases and Well-Known Types that are not handled yet)
- Oneof's are properly handled as sealed classes
- Specialized support to handle wrappers from the well-known types (e.g.
StringValue,BoolValue) as nullable primitives (String?,Boolean?, etc.) - Support for custom service/gRPC code generator
- Support for custom options
- Support for the
deprecatedprotobuf option when generating Kotlin code
- Kotlin/JS (
js)- Support for Kotlin/JS in pbandk is still experimental. Using pbandk from other Kotlin Multiplatform code on Kotlin/JS is mostly stable. Using pbandk and pbandk-generated code directly from Javascript/Typescript code is not well supported yet because of limitations in Kotlin's
@JsExportfunctionality.
- Support for Kotlin/JS in pbandk is still experimental. Using pbandk from other Kotlin Multiplatform code on Kotlin/JS is mostly stable. Using pbandk and pbandk-generated code directly from Javascript/Typescript code is not well supported yet because of limitations in Kotlin's
- Kotlin/Wasm (
wasmJs)- Support for Kotlin/Wasm in pbandk is still experimental. There is no support for using pbandk APIs and pbandk-generated code directly from Javascript code (or other Wasm host languages). The APIs can only be called from other Kotlin code that is being compiled for Kotlin/Wasm. Kotlin/Wasm doesn't currently support using classes or complex types as arguments for Kotlin functions that are exposed to the Wasm host environment. This makes it infeasible to expose pbandk APIs directly outside of Kotlin.
- Kotlin/Native - Windows (
mingwX64)- Windows is supported on a best-effort basis. The pbandk runtime is available on the
mingwX64platform. The official protobuf project does not yet support running the protobuf conformance tests on Windows, so we cannot fully validate pbandk's correctness when run on Windows. However, almost all of themingwX64pbandk code is shared with the other native platforms (Linux, macOS), so the passing conformance tests on those platforms still provide a pretty good guarantee of pbandk's correctness onmingwX64.
- Windows is supported on a best-effort basis. The pbandk runtime is available on the
- Specialized support for some of the less-common "well known types" (e.g.
FieldMask) - Access to the protobuf descriptor from generated code
- Code comments on generated code
Read below for more information and see the examples.
This project is currently in beta. It has the core set of protobuf features implemented and is being used in production. But it is still under active development and new versions might introduce backwards-incompatible changes to support new features or to improve the library's usability in Kotlin. Pull requests are welcome for any of the "Not Yet Implemented" features above.
This project follows semantic versioning. After v1.0.0 is released, future versions will preserve backwards compatibility.
The project currently has a single maintainer (@garyp) working on it in his spare time. Contributors who would like to become additional maintainers are highly welcome. Your contributions don't have to be in the form of code and could also be documentation improvements, issue triage, community outreach, etc.
For support or discussion relating to pbandk, please use the GitHub Discussions on this project. You can also find some of us in the #pbandk channel of the Kotlin Slack instance. Please drop in and say hi!
For the following addressbook.proto file:
syntax = "proto3";
package tutorial;
import "google/protobuf/timestamp.proto";
message Person {
string name = 1;
int32 id = 2;
string email = 3;
enum PhoneType {
MOBILE = 0;
HOME = 1;
WORK = 2;
}
message PhoneNumber {
string number = 1;
PhoneType type = 2;
}
repeated PhoneNumber phones = 4;
google.protobuf.Timestamp last_updated = 5;
}
message AddressBook {
repeated Person people = 1;
}The following file will be generated at tutorial/addressbook.kt:
@file:OptIn(pbandk.PublicForGeneratedCode::class)
package tutorial
@pbandk.Export
public data class Person(
val name: String = "",
val id: Int = 0,
val email: String = "",
val phones: List<tutorial.Person.PhoneNumber> = emptyList(),
val lastUpdated: pbandk.wkt.Timestamp? = null,
override val unknownFields: Map<Int, pbandk.UnknownField> = emptyMap()
) : pbandk.Message {
override operator fun plus(other: pbandk.Message?): tutorial.Person = protoMergeImpl(other)
override val descriptor: pbandk.MessageDescriptor<tutorial.Person> get() = Companion.descriptor
override val protoSize: Int by lazy { super.protoSize }
public companion object : pbandk.Message.Companion<tutorial.Person> {
public val defaultInstance: tutorial.Person by lazy { tutorial.Person() }
override fun decodeWith(u: pbandk.MessageDecoder): tutorial.Person = tutorial.Person.decodeWithImpl(u)
override val descriptors: pbandk.MessageDescriptor<tutorial.Person> = pbandk.MessageDescriptor(
fullName = "tutorial.Person",
messageClass = tutorial.Person::class,
messageCompanion = this,
fields = buildList(5) {
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "name",
number = 1,
type = pbandk.FieldDescriptor.Type.Primitive.String(),
jsonName = "name",
value = tutorial.Person::name
)
)
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "id",
number = 2,
type = pbandk.FieldDescriptor.Type.Primitive.Int32(),
jsonName = "id",
value = tutorial.Person::id
)
)
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "email",
number = 3,
type = pbandk.FieldDescriptor.Type.Primitive.String(),
jsonName = "email",
value = tutorial.Person::email
)
)
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "phones",
number = 4,
type = pbandk.FieldDescriptor.Type.Repeated<tutorial.Person.PhoneNumber>(valueType = pbandk.FieldDescriptor.Type.Message(messageCompanion = tutorial.Person.PhoneNumber.Companion)),
jsonName = "phones",
value = tutorial.Person::phones
)
)
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "last_updated",
number = 5,
type = pbandk.FieldDescriptor.Type.Message(messageCompanion = pbandk.wkt.Timestamp.Companion),
jsonName = "lastUpdated",
value = tutorial.Person::lastUpdated
)
)
}
)
}
public sealed class PhoneType(override val value: Int, override val name: String? = null) : pbandk.Message.Enum {
override fun equals(other: kotlin.Any?): Boolean = other is tutorial.Person.PhoneType && other.value == value
override fun hashCode(): Int = value.hashCode()
override fun toString(): String = "Person.PhoneType.${name ?: "UNRECOGNIZED"}(value=$value)"
public object MOBILE : PhoneType(0, "MOBILE")
public object HOME : PhoneType(1, "HOME")
public object WORK : PhoneType(2, "WORK")
public class UNRECOGNIZED(value: Int) : PhoneType(value)
public companion object : pbandk.Message.Enum.Companion<tutorial.Person.PhoneType> {
public val values: List<tutorial.Person.PhoneType> by lazy { listOf(MOBILE, HOME, WORK) }
override fun fromValue(value: Int): tutorial.Person.PhoneType = values.firstOrNull { it.value == value } ?: UNRECOGNIZED(value)
override fun fromName(name: String): tutorial.Person.PhoneType = values.firstOrNull { it.name == name } ?: throw IllegalArgumentException("No PhoneType with name: $name")
}
}
public data class PhoneNumber(
val number: String = "",
val type: tutorial.Person.PhoneType = tutorial.Person.PhoneType.fromValue(0),
override val unknownFields: Map<Int, pbandk.UnknownField> = emptyMap()
) : pbandk.Message {
override operator fun plus(other: pbandk.Message?): tutorial.Person.PhoneNumber = protoMergeImpl(other)
override val descriptor: MessageDescriptor<tutorial.Person.PhoneNumber> get() = Companion.descriptor
override val protoSize: Int by lazy { super.protoSize }
public companion object : pbandk.Message.Companion<tutorial.Person.PhoneNumber> {
public val defaultInstance: tutorial.Person.PhoneNumber by lazy { tutorial.Person.PhoneNumber() }
override fun decodeWith(u: pbandk.MessageDecoder): tutorial.Person.PhoneNumber = tutorial.Person.PhoneNumber.decodeWithImpl(u)
override val descriptor: pbandk.MessageDescriptor<tutorial.Person.PhoneNumber> = pbandk.MessageDescriptor(
fullName = "tutorial.Person.PhoneNumber",
messageClass = tutorial.Person.PhoneNumber::class,
messageCompanion = this,
fields = buildList(2) {
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "number",
number = 1,
type = pbandk.FieldDescriptor.Type.Primitive.String(),
jsonName = "number",
value = tutorial.Person.PhoneNumber::number
)
)
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "type",
number = 2,
type = pbandk.FieldDescriptor.Type.Enum(enumCompanion = tutorial.Person.PhoneType.Companion),
jsonName = "type",
value = tutorial.Person.PhoneNumber::type
)
)
}
)
}
}
}
@pbandk.Export
public data class AddressBook(
val people: List<tutorial.Person> = emptyList(),
override val unknownFields: Map<Int, pbandk.UnknownField> = emptyMap()
) : pbandk.Message {
override operator fun plus(other: pbandk.Message?): tutorial.AddressBook = protoMergeImpl(other)
override val descriptor: MessageDescriptor<tutorial.AddressBook> get() = Companion.descriptor
override val protoSize: Int by lazy { super.protoSize }
public companion object : pbandk.Message.Companion<tutorial.AddressBook> {
public val defaultInstance: tutorial.AddressBook by lazy { tutorial.AddressBook() }
override fun decodeWith(u: pbandk.MessageDecoder): tutorial.AddressBook = tutorial.AddressBook.decodeWithImpl(u)
override val descriptor: pbandk.MessageDescriptor<tutorial.AddressBook> = pbandk.MessageDescriptor(
fullName = "tutorial.AddressBook",
messageClass = tutorial.AddressBook::class,
messageCompanion = this,
fields = buildList(1) {
add(
pbandk.FieldDescriptor(
messageDescriptor = this@Companion::descriptor,
name = "people",
number = 1,
type = pbandk.FieldDescriptor.Type.Repeated<tutorial.Person>(valueType = pbandk.FieldDescriptor.Type.Message(messageCompanion = tutorial.Person.Companion)),
jsonName = "people",
value = tutorial.AddressBook::people
)
)
}
)
}
}
@pbandk.Export
@pbandk.JsName("orDefaultForPerson")
public fun Person?.orDefault(): tutorial.Person = this ?: Person.defaultInstance
@pbandk.Export
@pbandk.JsName("orDefaultForPersonPhoneNumber")
public fun Person.PhoneNumber?.orDefault(): tutorial.Person.PhoneNumber = this ?: Person.PhoneNumber.defaultInstance
@pbandk.Export
@pbandk.JsName("orDefaultForAddressBook")
public fun AddressBook?.orDefault(): tutorial.AddressBook = this ?: AddressBook.defaultInstance
// Omitted multiple supporting private extension methodsTo see a full version of the file, see here. See the "Generated Code" section below under "Usage" for more details.
Pbandk's code generator leverages protoc. Download the latest
protoc and make sure protoc is on the PATH.
Then download the latest released protoc-gen-pbandk self-executing jar
file (if you're using a SNAPSHOT build of pbandk, you might want to instead download the latest SNAPSHOT version of protoc-gen-pbandk-jvm-*-jvm8.jar),
rename it to protoc-gen-pbandk, make the file executable (chmod +x protoc-gen-pbandk), and make sure it is on the PATH. To generate code from
sample.proto and put the generated code in src/main/kotlin, run:
protoc --pbandk_out=src/main/kotlin sample.proto
The file is generated as sample.kt in the subdirectories specified by the package. Like other X_out arguments,
comma-separated options can be added to --pbandk_out before the colon and out dir path:
-
To explicitly set the Kotlin package to
my.pkg, use thekotlin_packageoption like so:protoc --pbandk_out=kotlin_package=my.pkg:src/main/kotlin sample.proto -
If you have multiple proto packages, you can map them using
kotlin_package_mappingoption like so:protoc --pbandk_out=kotlin_package_mapping="simple.package->new.package;foo.bar.*->my.foo.bar.*":src/main/kotlin sample.proto -
By default all generated classes have
publicvisibility. To change the visibility tointernal, use thevisibilityoption like so:protoc --pbandk_out=visibility=internal:src/main/kotlin sample.proto -
To log debug logs during generation,
log=debugcan be set as well.
Multiple options can be added to a single --pbandk_out argument by separating them with commas.
In addition to running protoc manually, the
Protobuf Plugin for Gradle can be used. See
this example to see how.
The self-executing jar file doesn't work on Windows. Also protoc doesn't support finding
protoc-gen-pbandk.bat on the PATH. So it has to be specified explicitly as a plugin. Thus on
Windows you will first need to build protoc-gen-pbandk locally:
./gradlew :protoc-gen-pbandk:protoc-gen-pbandk-jvm:installDist
And then provide the full path to protoc:
protoc \
--pbandk_out=src/main/kotlin \
--plugin=protoc-gen-pbandk=/path/to/pbandk/protoc-gen-pbandk/jvm/build/install/protoc-gen-pbandk/bin/protoc-gen-pbandk.bat \
sample.proto
Pbandk's runtime library provides a Kotlin layer over the preferred Protobuf library for each platform. The libraries are present on Maven Central. Using Gradle:
repositories {
// This repository is only needed if using a SNAPSHOT version of pbandk
maven { url "https://s01.oss.sonatype.org/content/repositories/snapshots/" }
mavenCentral()
}
dependencies {
// Can be used from the `common` sourceset in a Kotlin Multiplatform project,
// or from platform-specific JVM, Android, JS, or Native sourcesets/projects.
implementation("pro.streem.pbandk:pbandk-runtime:0.16.1-SNAPSHOT")
}
Pbandk does not generate gRPC code itself, but offers a pbandk.gen.ServiceGenerator interface in
the protoc-gen-pbandk-lib-jvm project with a single method that can be implemented to generate the
code.
To do this, first depend on the project but it will only be needed at compile time because it's already there at runtime:
dependencies {
compileOnly("pro.streem.pbandk:protoc-gen-pbandk-lib:0.16.1-SNAPSHOT")
}
Then, the kotlin_service_gen option can be given to protoc to use the generator. The option is a path-separated
collection of JAR files to put on the classpath. It can end with a pipe (i.e. |) following by the fully-qualified
class name of the implementation of the ServiceGenerator to use. If the last part is not present, it will use the
ServiceLoader mechanism to find the first implementation to use. For example, to gen with my.Generator from
gen.jar, it might look like:
protoc --pbandk_out=kotlin_service_gen=gen.jar|my.Generator,kotlin_package=my.pkg:src/main/kotlin some.proto
For more details, see the custom-service-gen example.
The package is either the kotlin_package plugin option, the java_package protobuf option or the package set in the message. If the google.protobuf
package is referenced, it is assumed to be a well-known type and is changed to reference pbandk.wkt.
Each Protobuf message extends pbandk.Message and has an encodeToByteArray method to encode the message with the
Protobuf binary encoding into a ByteArray. The companion object of every message has a decodeFromByteArray method: it
accepts a ByteArray and returns an instance of the class. Each platform also provides additional encodeTo* and
decodeFrom* methods that are platform-specific. For example, the JVM provides encodeToStream and decodeFromStream
methods that operate on Java's OutputStream and InputStream, respectively. There are also encodeToJsonString and
decodeFromJsonString methods that use the Protobuf JSON encoding to encode/decode the message into a string.
Messages are immutable Kotlin data classes. This means they automatically implement hashCode, equals, and
toString. Each class has an unknownFields map which contains information about extra fields the decoder didn't
recognize. If there are values in this map, they will be encoded on output. The MessageDecoder instances have a
constructor option to discard unknown fields when reading.
For proto3, the only nullable fields are messages, oneof fields, and optional fields. Other values have defaults. For
proto2, optional fields are nullable and defaulted as such. Types are basically the same as they would be in Java.
However, bytes fields actually use a pbandk.ByteArr class which is a simple wrapper around ByteArray. This was done
because Kotlin does not handle array fields in data classes predictably and it wasn't worth overriding equals and
hashCode every time.
Regardless of optimize_for options, the generated code is always the same. Each message has a protoSize field that
lazily calculates the size of the message when first invoked. Also, each message has the plus operator defined which
follows protobuf merge semantics.
Oneof fields are generated as nested classes of a common sealed base class. Each oneof inner field is a class that wraps a single value.
The parent message also contains a nullable field for every oneof inner field. This field resolves to the oneof inner field's value when the oneof is set to that inner field. Otherwise it resolves to null.
Enum fields are generated as sealed classes with a nested object for each known enum value, and a
Unrecognized nested class to hold unknown values. This is preferred over traditional Kotlin enum classes
because enums in protobuf are open ended and may not be one of the specific known values. Traditional
enum classes would not be able to capture this state, and using sealed classes this way requires the
user to do explicit checks for the Unrecognized value during exhaustive when clauses.
Each enum object contains a value field with the numeric value of that enum, and a name field
with the string value of that enum. Developers should use the fromValue and fromName methods
present on the companion object of the sealed class to map from a numeric or string value,
respectively, to the corresponding enum object.
The values field on the companion object of the sealed class contains a list of all known enum
values.
repeated fields are normal Lists. Developers should make no assumptions about which list implementation is used.
maps are represented by Kotlin Maps. In proto2, due to how map entries are serialized, both the key and the value
are considered nullable.
Well known types (i.e. those in the google/protobuf imports) are shipped with the runtime under the pbandk.wkt package.
Specialized support is provided to map the types defined in google/protobuf/wrappers.proto into Kotlin nullable primitives (e.g. String? for google.protobuf.StringValue, Int? for google.protobuf.Int32Value, etc.).
Services can be handled with a custom service generator. See the "Service Code Generation" section above and the custom-service-gen example.
The project is built with Gradle and has several sub projects. They are:
conformance/js- Conformance test runner for Kotlin/JSconformance/jvm- Conformance test runner for Kotlin/JVMconformance/native- Conformance test runner for Kotlin/Nativeconformance/wasmJs- Conformance test runner for Kotlin/Wasmconformance/lib- Common multiplatform code for conformance testsprotoc-gen-pbandk/jvm- Kotlin/JVM implementation of the code generator (can generate code for any platform, but requires JVM to run)protoc-gen-pbandk/lib- Multiplatform code (only Kotlin/JVM supported at the moment) for the code generator andServiceGeneratorlibraryprotos- Protobuf definitions of the Well-Known Types, packaged as a separate project for compatibility with the Protobuf Gradle Pluginruntime- Kotlin Multiplatform library for runtime Protobuf supporttest-types- Protobuf definitions and generated code for protobuf messages used in pbandk unit tests
To generate the protoc-gen-pbandk distribution, run:
./gradlew :protoc-gen-pbandk:protoc-gen-pbandk-jvm:assembleDist
If you want to make changes to pbandk, and immediately test these changes in your separate project,
first install the generator locally:
./gradlew :protoc-gen-pbandk:protoc-gen-pbandk-jvm:installDist
This puts the files in the build/install folder. Then you need to tell protoc where to find this plugin file.
For example:
protoc \
--plugin=protoc-gen-pbandk=/path/to/pbandk/protoc-gen-pbandk/jvm/build/install/protoc-gen-pbandk/bin/protoc-gen-pbandk \
--pbandk_out=src/main/kotlin \
src/main/proto/*.proto
This will generate kotlin files for the specified *.proto files, without needing to publish first.
To build the runtime library for both JS and the JVM, run:
./gradlew :pbandk-runtime:assemble
If any changes are made to the generated code that is output by protoc-gen-pbandk, then the
well-known types (and other proto types used by pbandk) need to be re-generated using the updated
protoc-gen-pbandk binary:
./gradlew generateProtos
Important: If making changes in both the :protoc-gen-pbandk:protoc-gen-pbandk-lib and :pbandk-runtime projects at
the same time, then it's likely the :pbandk-runtime:generateWellKnownTypeProtos task will fail to compile. To work
around this, stash the changes in the :pbandk-runtime project, run the generateWellKnownTypeProtos task with only
the :protoc-gen-pbandk:protoc-gen-pbandk-lib changes, and then unstash the :pbandk-runtime changes and rerun the
generateWellKnownTypeProtos task.
To run conformance tests, the conformance-test-runner must be built (does not work on Windows).
git clone -b v28.0 --depth 1 --recurse-submodules --shallow-submodules https://github.com/protocolbuffers/protobuf.git
cd protobuf
cmake -S . -B build -DCMAKE_CXX_STANDARD=14 -Dprotobuf_BUILD_CONFORMANCE=ON -Dprotobuf_BUILD_TESTS=OFF -Dprotobuf_BUILD_LIBUPB=OFF
cmake --build build --parallel=10
You should now have a conformance_test_runner available in protobuf/build directory. Test it by running ./conformance_test_runner --help
Set the CONF_TEST_PATH environment variable (used to run the tests below) with:
export CONF_TEST_PATH="$(pwd)/conformance_test_runner"
Now, back in pbandk, build all conformance sub-projects via:
./gradlew :conformance:conformance-lib:assemble \
:conformance:conformance-jvm:installDist \
:conformance:conformance-native:build
You are now ready to run the conformance tests. Make sure CONF_TEST_PATH environment variable is set to path/to/protobuf/build/conformance_test_runner (see above).
Then, from the root directory:
./conformance/test-conformance.sh
Note that by default, the test-conformance.sh script will run the conformance test for jvm, js, wasmJs, and linux. This will fail when running them on MacOS due to missing linux binaries. So in that case, run the tests for each platform individually:
./conformance/test-conformance.sh jvm
./conformance/test-conformance.sh js
./conformance/test-conformance.sh wasmJs
./conformance/test-conformance.sh macos
Releases are handled automatically via CI once the git tag is created.
Setup a couple shell variables to simplify the rest of the commands below:
export VERSION="0.9.0"
export NEXT_VERSION="0.9.1"To create a new release:
- Update
CHANGELOG.md: add a date for the release version, and update the release version's GitHub compare link with a tag instead ofHEAD.- Note: if you are releasing a pre-release version (alpha, beta, rc) then you don't need to update
CHANGELOG.md
- Note: if you are releasing a pre-release version (alpha, beta, rc) then you don't need to update
- Update the pbandk version number in
gradle.properties,README.md, andexamples/*/build.gradle.ktsto remove theSNAPSHOTsuffix. For example, if the current version is0.9.0-SNAPSHOT, then update it to be0.9.0. - Comment out the note about the stable version of the documentation that is at the top of
README.mdand update it to point at the new version. Also update the link toprotoc-gen-pbandk-jvminREADME.mdto point at the new version. - Commit the change. E.g.:
git commit -m "Bump to ${VERSION}" -a. - Tag the new version. E.g.:
git tag -a -m "See https://github.com/streem/pbandk/blob/v${VERSION}/CHANGELOG.md" "v${VERSION}".
Then prepare the repository for development of the next version:
- Update
CHANGELOG.md: add a section forNEXT_VERSIONthat will follow the released version (e.g. if releasing0.9.0then add a section for0.9.1).- Note: if you are releasing a pre-release version (alpha, beta, rc) then you don't need to update
CHANGELOG.md
- Note: if you are releasing a pre-release version (alpha, beta, rc) then you don't need to update
- Update the pbandk version number in
gradle.properties,README.md, andexamples/*/build.gradle.ktsto${NEXT_VERSION}-SNAPSHOT. For example,0.9.1-SNAPSHOT. - Uncomment the note about the stable version of the documentation that is at the top of
README.md. - Commit the change. E.g.:
git commit -m "Bump to ${NEXT_VERSION}-SNAPSHOT" -a.
GitHub will build and publish the new release once it sees the new tag:
- Push the changes to GitHub:
git push origin --follow-tags master. - Wait for CI to notice the new tag, build it, and upload it to Maven Central.
- Create a new release on GitHub. Use the contents of the tag description as the release description. E.g.:
gh release create "v${VERSION}" -F <(git tag -l --format='%(contents)' "v${VERSION}").
This repository was originally forked from https://github.com/cretz/pb-and-k. Many thanks to https://github.com/cretz for creating this library and building the initial feature set.
pbandk uses its own pure-Kotlin protobuf implementation that is heavily based on the Google Protobuf Java library.