Create Azure Functions with Rust!
Branch: master
Clone or download
peterhuene Merge pull request #165 from peterhuene/eventgrid-bindings
Implement Event Grid trigger binding.
Latest commit 070ab54 Feb 20, 2019

Azure Functions for Rust CircleCI branch Dependabot Status license

This is an early-stage framework for implementing Azure Functions in Rust.


Although the maintainer of this project is a Microsoft employee, this project is not an officially recognized Microsoft product and is not an endorsement of any future product offering from Microsoft.

Microsoft and Azure are registered trademarks of Microsoft Corporation.


A simple HTTP-triggered Azure Function:

use azure_functions::bindings::{HttpRequest, HttpResponse};
use azure_functions::func;

pub fn greet(req: &HttpRequest) -> HttpResponse {
    // Log the message with the Azure Functions Host
    info!("Request: {:?}", req);

        "Hello from Rust, {}!\n",
        req.query_params().get("name").map_or("stranger", |x| x)

See the examples directory for the complete list of examples.


Documentation for the latest published version.

Getting Started

Install CMake

The azure-functions crate has a dependency on the grpcio crate that uses CMake to build and link against the gRPC native library.

CMake must be installed and on the PATH to be able to use Azure Functions for Rust.


Install CMake from the Windows installer.


The easiest way to install CMake on macOS is with Homebrew:

$ brew install cmake


Use your distro's package manager to install a cmake (or similar) package.

For example on Debian/Ubuntu:

$ apt-get install cmake

Install the Azure Functions Core Tools

Install version 2 or higher of the Azure Functions Core Tools.

If you are on Windows, you must add %ProgramFiles%\nodejs\node_modules\azure-functions-core-tools\bin (where func.exe is located) to the PATH environment variable.

Installing the Azure Functions for Rust SDK

Install the Azure Functions for Rust SDK using cargo install:

$ cargo install azure-functions-sdk

This installs a new cargo command named func that can be used to create and run new Azure Functions applications.

Creating a new Azure Functions application

Use the cargo func new-app command to create a new Azure Functions application:

$ cargo func new-app hello

This will create a new application in the ./hello directory with a module named functions where the exported Azure Functions are expected to be placed.

Adding a simple HTTP-triggered application

Use the cargo func new command to create a new HTTP-triggered Azure Function named hello:

$ cargo func new http -n hello

The source for the function will be in src/functions/

Building the Azure Functions application

To build your Azure Functions application, just use cargo build:

$ cargo build

Running the Azure Functions application

To build and run your Azure Functions application, use cargo func run:

$ cargo func run

The cargo func run command builds and runs your application locally using the Azure Function Host that was installed by the Azure Functions Core Tools.

By default, the host will be configured to listen on

For the hello function added previously, it can be invoked from http://localhost:8080/api/hello.

Debugging the Azure Functions application

The easiest way to debug the Azure Functions application is to use Visual Studio Code with the CodeLLDB extension.

Copy the example launch.json and tasks.json files to the .vscode directory inside the root of your project.

This will enable a Debug launch configuration that will build and run your application locally before attaching the lldb debugger to the Rust worker process.

Deploying the Azure Functions application

In the future, there will be a cargo func deploy command to deploy the Azure Functions application directly to Azure.

Until that time, you must manually build and push the Docker image to a repository that can be accessed by Azure.

Note: this requires Docker that is at least 18.06 for the experimental BuildKit support.

To enable the BuildKit support, set the DOCKER_BUILDKIT environment variable to 1 before running docker build.

Start by building your image with docker build -t $TAG_NAME .:

$ docker build -t $TAG_NAME .

Where $TAG_NAME is the tag name to use (e.g. username/my-functions-app).

Use docker push to push the image to a repository that is accessible to Azure Functions.

$ docker push $TAG_NAME

Create the Function App in Azure using the "Linux (Preview)" OS. Under the "Publish" setting, select "Docker Image".

From the "Configure Container" section, select the repository and enter the image you pushed.

That's it! Once the Functions App starts in Azure, you should be able to view the functions and run them.

Azure Functions Bindings

Azure Functions supports a wide variety of input and output bindings that can be used by a function.

In a language like C#, parameters can be annotated with attributes describing how the parameters are bound.

Rust does not support attributes on parameters, so the #[binding] attribute is applied on the function with a name that matches the parameter's identifier. The arguments to the attribute depend on the binding type.

The #[func] attribute is used to turn an ordinary Rust function into an Azure Function. It works by examining the parameters and return type to the function and automatically mapping them to corresponding bindings.

The current list of supported bindings:

Rust Type Azure Functions Binding Direction
azure_functions::bindings::Blob Input and Ouput Blob in, inout, out
azure_functions::bindings::BlobTrigger Blob Trigger in, inout
azure_functions::bindings::EventGridEvent Event Grid Event in
azure_functions::bindings::HttpRequest HTTP Trigger in
azure_functions::bindings::HttpResponse Output HTTP Response out
azure_functions::bindings::QueueTrigger Queue Trigger in
azure_functions::bindings::QueueMessage Output Queue Message out
azure_functions::bindings::Table Input and Ouput Table in, out
azure_functions::bindings::TimerInfo Timer Trigger in
azure_functions::Context* Invocation Context n/a

*Note: the Context binding is not an Azure Functions binding; it is used to pass information about the function being invoked.

More bindings will be implemented in the future, including support for retreiving data from custom bindings.

Bindings in Rust

Azure Functions for Rust automatically infers the direction of bindings depending on how the binding is used in a function's declaration:

  • Parameters passed by immutable reference &T, where T is a trigger or input binding type, are inferred to be bindings with an in direction.

    pub fn example(..., blob: &Blob) {
  • Parameters passed by mutable reference &mut T, where T is a trigger or input binding type that supports the inout direction, are inferred to be bindings with an inout direction. Note: inout direction bindings are currently not implemented for languages other than C#. See this issue regarding this problem with the Azure Functions Host.

    pub fn example(..., blob: &mut Blob) {
  • Functions that return a type T, where T is an output binding type, or a tuple of output binding types, are inferred to be bindings with an out direction. Functions may also return Option<T> for any output binding type T; a None value will skip outputting a value.

    pub fn example(...) -> Blob {
    pub fn example(...) -> Option<Blob> {
    pub fn example(...) -> (HttpResponse, Option<Blob>) {

    For functions that return a single output binding type, the binding has a special name of $return and is treated as the "return value" of the function.

    For functions that return a tuple of output binding types, the first value of the tuple has the binding name of $return and is treated as the "return value" of the function. The remaining values have binding names output1, output2, ..., output(N-1) where N is the number of types in the tuple, and are treated as output bindings only.

    Unit tuples () can be used in a tuple to "skip" a binding:

    pub fn example(...) -> ((), Blob) {

    For the above example, there is no $return binding and the Azure Function "returns" no value. Instead, a single output binding named output1 is used.


Cloning the Repository

This repository uses a git submodule for defining the Azure Functions Language Worker Protocol.

Use --recurse-submodules when cloning this repository:

$ git clone --recurse-submodules

Repository Layout

This repository is split into multiple Rust crates:

  • azure-functions - The azure-functions crate that defines the types and functions that are used when writing Azure Functions with Rust.
  • azure-functions-codegen - The azure-functions-codegen crate that defines the procedural macros that are used when writing Azure Functions with Rust.
  • azure-functions-sdk - The azure-functions-sdk crate that implements the cargo func command.
  • azure-functions-shared - The azure-functions-shared crate that defines types and functions that are shared between the azure-functions-codegen and azure-functions crates.
  • azure-functions-shared-codegen - The azure-functions-shared-codegen crate that defines the procedural macros used by the shared azure-functions-shared crate.
  • examples - The directory containing example Azure Functions.


Build at the root of the repository to build both the azure-functions-codegen and the azure-functions libraries using cargo build:

$ cargo build

Running tests

Use cargo test to run the tests:

$ cargo test