WARNING: Breaking changes are likely to happen until v1.0 is released
Arri RPC is a code-first RPC framework. Type-safe clients get generated directly from your server code meaning you never need to manually write another client again.
Go here to get started with a Typescript server. More server implementations to come.
https://github.com/user-attachments/assets/15cf68a4-871e-4e7d-b5fc-25dcd1760fc1
When I have time I would like to add more languages to this list. Currently I have the following languages on my shortlist for potential server implementations:
See this guide for information on how to implement your own Arri server
Below are the language client generators that are planned to have first party support. This chart tracks the current progress on implementations for these clients. For those interested in creating their own generators, see this guide.
Language | HTTP | SSE |
---|---|---|
Typescript | ✅ | ✅ |
Dart | ✅ | ✅ |
Rust | ✅ | ✅ |
Kotlin | ✅ | ✅ |
Swift | ✅ | ✅ |
Go | ||
Python |
✅ completed
🚧 in progress
Even though Arri focuses primarily on a code-first approach it allow you to manually create app definitions in the event that you have a server implementation that isn't supported. Once your app definition is created you simply need to point the CLI to the app definition file. This file can be can be a typescript file, JSON file, or a JSON http endpoint.
arri codegen ./AppDefinition.ts
arri codegen ./AppDefinition.json
arri codegen https://myapi.com/rpcs/__definition # must accept a GET request
Before running this command. Make sure you have an arri config created already.
Example Config:
// arri.config.ts
import { defineConfig, generators } from "arri";
export default defineConfig({
generators: [
generators.dartClient({
// options
}),
generators.kotlinClient({
// options
}),
generators.typescriptClient({
// options
}),
],
});
Arri comes with some useful helpers that reduces the boilerplate of manually creating a JSON definition file. Additionally the validators created with Arri Schema can be used throughout your app.
// AppDefinition.ts
import { createAppDefinition } from "arri";
import { a } from "@arrirpc/schema";
const HelloParams = a.object("HelloParams", {
message: a.string(),
});
const HelloResponse = a.object("HelloResponse", {
message: a.string(),
});
export default createAppDefinition({
procedures: {
sayHello: {
transport: "http",
method: "post",
path: "/say-hello",
params: HelloParams,
response: HelloResponse,
},
},
});
Additionally if you only need cross language types, you can skip defining procedures all together and just pass in models to the helper.
// AppDefinition.ts
import { createAppDefinition } from "arri";
import { a } from "@arrirpc/schema";
const HelloParams = a.object("HelloParams", {
message: a.string(),
});
const HelloResponse = a.object("HelloResponse", {
message: a.string(),
});
export default createAppDefinition({
definitions: {
HelloParams,
HelloResponse,
},
});
Now arri codegen ./AppDefinition.ts
will only generate types for each client defined in the arri config.
JSON app definitions are something that would normally be automatically generated by an implementation of ARRI-RPC. Manually creating a JSON app definition is more terse and more subject to human error than the typescript alternative.
{
"schemaVersion": "<current-schema-version>",
"procedures": {
"sayHello": {
"transport": "http",
"method": "get",
"path": "/say-hello",
"params": "HelloParams",
"response": "HelloResponse"
}
},
"definitions": {
"HelloParams": {
"properties": {
"message": {
"type": "string",
"metadata": {}
}
},
"metadata": {
"id": "HelloParams",
"metadata": {}
}
},
"HelloResponse": {
"properties": {
"message": {
"type": "string",
"metadata": {}
}
},
"metadata": {
"id": "HelloResponse"
}
}
}
}
Contributions are welcome!
Please read the contribution guide which will guide you through the entire workflow of how to build the source code, how to run the tests, and how to contribute changes to the Arri RPC codebase. Also feel free to reach out on discord if you have any other additional questions.