The Cargo Book

Cargo is the Rust package manager. Cargo downloads your Rust package’s dependencies, compiles your packages, makes distributable packages, and uploads them to crates.io, the Rust community’s package registry. You can contribute to this book on GitHub.

Sections

Getting Started

To get started with Cargo, install Cargo (and Rust) and set up your first crate.

Cargo Guide

The guide will give you all you need to know about how to use Cargo to develop Rust packages.

Cargo Reference

The reference covers the details of various areas of Cargo.

Frequently Asked Questions

Getting Started

To get started with Cargo, install Cargo (and Rust) and set up your first crate.

• Installation
• First steps with Cargo

Installation

Install Rust and Cargo

The easiest way to get Cargo is to install the current stable release of Rust by using rustup. Installing Rust using rustup will also install cargo.

On Linux and macOS systems, this is done as follows:

$ curl https://sh.rustup.rs -sSf | sh

It will download a script, and start the installation. If everything goes well, you’ll see this appear:

Rust is installed now. Great!

On Windows, download and run rustup-init.exe. It will start the installation in a console and present the above message on success.

After this, you can use the rustup command to also install beta or nightly channels for Rust and Cargo.

For other installation options and information, visit the install page of the Rust website.

Build and Install Cargo from Source

Alternatively, you can build Cargo from source.

First Steps with Cargo

To start a new package with Cargo, use cargo new:

$ cargo new hello_world

Cargo defaults to --bin to make a binary program. To make a library, we'd pass --lib.

Let’s check out what Cargo has generated for us:

$ cd hello_world
$ tree .
.
├── Cargo.toml
└── src
    └── main.rs

1 directory, 2 files

This is all we need to get started. First, let’s check out Cargo.toml:

[package]
name = "hello_world"
version = "0.1.0"
authors = ["Your Name <you@example.com>"]
edition = "2018"

[dependencies]

This is called a manifest, and it contains all of the metadata that Cargo needs to compile your package.

Here’s what’s in src/main.rs:

fn main() {
    println!("Hello, world!");
}

Cargo generated a “hello world” for us. Let’s compile it:

$ cargo build
   Compiling hello_world v0.1.0 (file:///path/to/package/hello_world)

And then run it:

$ ./target/debug/hello_world
Hello, world!

We can also use cargo run to compile and then run it, all in one step:

$ cargo run
     Fresh hello_world v0.1.0 (file:///path/to/package/hello_world)
   Running `target/hello_world`
Hello, world!

Going further

For more details on using Cargo, check out the Cargo Guide

Cargo Guide

This guide will give you all that you need to know about how to use Cargo to develop Rust packages.

• Why Cargo Exists
• Creating a New Package
• Working on an Existing Cargo Package
• Dependencies
• Package Layout
• Cargo.toml vs Cargo.lock
• Tests
• Continuous Integration
• Cargo Home
• Build Cache

Why Cargo Exists

Cargo is a tool that allows Rust packages to declare their various dependencies and ensure that you’ll always get a repeatable build.

To accomplish this goal, Cargo does four things:

• Introduces two metadata files with various bits of package information.
• Fetches and builds your package’s dependencies.
• Invokes rustc or another build tool with the correct parameters to build your package.
• Introduces conventions to make working with Rust packages easier.

Creating a New Package

To start a new package with Cargo, use cargo new:

$ cargo new hello_world --bin

We’re passing --bin because we’re making a binary program: if we were making a library, we’d pass --lib. This also initializes a new git repository by default. If you don't want it to do that, pass --vcs none.

Let’s check out what Cargo has generated for us:

$ cd hello_world
$ tree .
.
├── Cargo.toml
└── src
    └── main.rs

1 directory, 2 files

Let’s take a closer look at Cargo.toml:

[package]
name = "hello_world"
version = "0.1.0"
authors = ["Your Name <you@example.com>"]
edition = "2018"

[dependencies]

This is called a manifest, and it contains all of the metadata that Cargo needs to compile your package.

Here’s what’s in src/main.rs:

fn main() {
    println!("Hello, world!");
}

Cargo generated a “hello world” for us. Let’s compile it:

$ cargo build
   Compiling hello_world v0.1.0 (file:///path/to/package/hello_world)

And then run it:

$ ./target/debug/hello_world
Hello, world!

We can also use cargo run to compile and then run it, all in one step (You won't see the Compiling line if you have not made any changes since you last compiled):

$ cargo run
   Compiling hello_world v0.1.0 (file:///path/to/package/hello_world)
     Running `target/debug/hello_world`
Hello, world!

You’ll now notice a new file, Cargo.lock. It contains information about our dependencies. Since we don’t have any yet, it’s not very interesting.

Once you’re ready for release, you can use cargo build --release to compile your files with optimizations turned on:

$ cargo build --release
   Compiling hello_world v0.1.0 (file:///path/to/package/hello_world)

cargo build --release puts the resulting binary in target/release instead of target/debug.

Compiling in debug mode is the default for development-- compilation time is shorter since the compiler doesn't do optimizations, but the code will run slower. Release mode takes longer to compile, but the code will run faster.

Working on an Existing Cargo Package

If you download an existing package that uses Cargo, it’s really easy to get going.

First, get the package from somewhere. In this example, we’ll use rand cloned from its repository on GitHub:

$ git clone https://github.com/rust-lang-nursery/rand.git
$ cd rand

To build, use cargo build:

$ cargo build
   Compiling rand v0.1.0 (file:///path/to/package/rand)

This will fetch all of the dependencies and then build them, along with the package.

Dependencies

crates.io is the Rust community's central package registry that serves as a location to discover and download packages. cargo is configured to use it by default to find requested packages.

To depend on a library hosted on crates.io, add it to your Cargo.toml.

Adding a dependency

If your Cargo.toml doesn't already have a [dependencies] section, add that, then list the crate name and version that you would like to use. This example adds a dependency of the time crate:

[dependencies]
time = "0.1.12"

The version string is a semver version requirement. The specifying dependencies docs have more information about the options you have here.

If we also wanted to add a dependency on the regex crate, we would not need to add [dependencies] for each crate listed. Here's what your whole Cargo.toml file would look like with dependencies on the time and regex crates:

[package]
name = "hello_world"
version = "0.1.0"
authors = ["Your Name <you@example.com>"]
edition = "2018"

[dependencies]
time = "0.1.12"
regex = "0.1.41"

Re-run cargo build, and Cargo will fetch the new dependencies and all of their dependencies, compile them all, and update the Cargo.lock:

$ cargo build
      Updating crates.io index
   Downloading memchr v0.1.5
   Downloading libc v0.1.10
   Downloading regex-syntax v0.2.1
   Downloading memchr v0.1.5
   Downloading aho-corasick v0.3.0
   Downloading regex v0.1.41
     Compiling memchr v0.1.5
     Compiling libc v0.1.10
     Compiling regex-syntax v0.2.1
     Compiling memchr v0.1.5
     Compiling aho-corasick v0.3.0
     Compiling regex v0.1.41
     Compiling hello_world v0.1.0 (file:///path/to/package/hello_world)

Our Cargo.lock contains the exact information about which revision of all of these dependencies we used.

Now, if regex gets updated, we will still build with the same revision until we choose to cargo update.

You can now use the regex library in main.rs.

use regex::Regex;

fn main() {
    let re = Regex::new(r"^\d{4}-\d{2}-\d{2}$").unwrap();
    println!("Did our date match? {}", re.is_match("2014-01-01"));
}

Running it will show:

$ cargo run
   Running `target/hello_world`
Did our date match? true

Package Layout

Cargo uses conventions for file placement to make it easy to dive into a new Cargo package:

.
├── Cargo.lock
├── Cargo.toml
├── benches
│   └── large-input.rs
├── examples
│   └── simple.rs
├── src
│   ├── bin
│   │   └── another_executable.rs
│   ├── lib.rs
│   └── main.rs
└── tests
    └── some-integration-tests.rs

• Cargo.toml and Cargo.lock are stored in the root of your package (package root).
• Source code goes in the src directory.
• The default library file is src/lib.rs.
• The default executable file is src/main.rs.
• Other executables can be placed in src/bin/*.rs.
• Integration tests go in the tests directory (unit tests go in each file they're testing).
• Examples go in the examples directory.
• Benchmarks go in the benches directory.

These are explained in more detail in the manifest description.

Cargo.toml vs Cargo.lock

Cargo.toml and Cargo.lock serve two different purposes. Before we talk about them, here’s a summary:

• Cargo.toml is about describing your dependencies in a broad sense, and is written by you.
• Cargo.lock contains exact information about your dependencies. It is maintained by Cargo and should not be manually edited.

If you’re building a non-end product, such as a rust library that other rust packages will depend on, put Cargo.lock in your .gitignore. If you’re building an end product, which are executable like command-line tool or an application, or a system library with crate-type of staticlib or cdylib, check Cargo.lock into git. If you're curious about why that is, see "Why do binaries have Cargo.lock in version control, but not libraries?" in the FAQ.

Let’s dig in a little bit more.

Cargo.toml is a manifest file in which we can specify a bunch of different metadata about our package. For example, we can say that we depend on another package:

[package]
name = "hello_world"
version = "0.1.0"
authors = ["Your Name <you@example.com>"]

[dependencies]
rand = { git = "https://github.com/rust-lang-nursery/rand.git" }

This package has a single dependency, on the rand library. We’ve stated in this case that we’re relying on a particular Git repository that lives on GitHub. Since we haven’t specified any other information, Cargo assumes that we intend to use the latest commit on the master branch to build our package.

Sound good? Well, there’s one problem: If you build this package today, and then you send a copy to me, and I build this package tomorrow, something bad could happen. There could be more commits to rand in the meantime, and my build would include new commits while yours would not. Therefore, we would get different builds. This would be bad because we want reproducible builds.

We could fix this problem by putting a rev line in our Cargo.toml:

[dependencies]
rand = { git = "https://github.com/rust-lang-nursery/rand.git", rev = "9f35b8e" }

Now our builds will be the same. But there’s a big drawback: now we have to manually think about SHA-1s every time we want to update our library. This is both tedious and error prone.

Enter the Cargo.lock. Because of its existence, we don’t need to manually keep track of the exact revisions: Cargo will do it for us. When we have a manifest like this:

[package]
name = "hello_world"
version = "0.1.0"
authors = ["Your Name <you@example.com>"]

[dependencies]
rand = { git = "https://github.com/rust-lang-nursery/rand.git" }

Cargo will take the latest commit and write that information out into our Cargo.lock when we build for the first time. That file will look like this:

[[package]]
name = "hello_world"
version = "0.1.0"
dependencies = [
 "rand 0.1.0 (git+https://github.com/rust-lang-nursery/rand.git#9f35b8e439eeedd60b9414c58f389bdc6a3284f9)",
]

[[package]]
name = "rand"
version = "0.1.0"
source = "git+https://github.com/rust-lang-nursery/rand.git#9f35b8e439eeedd60b9414c58f389bdc6a3284f9"

You can see that there’s a lot more information here, including the exact revision we used to build. Now when you give your package to someone else, they’ll use the exact same SHA, even though we didn’t specify it in our Cargo.toml.

When we’re ready to opt in to a new version of the library, Cargo can re-calculate the dependencies and update things for us:

$ cargo update           # updates all dependencies
$ cargo update -p rand   # updates just “rand”

This will write out a new Cargo.lock with the new version information. Note that the argument to cargo update is actually a Package ID Specification and rand is just a short specification.

Tests

Cargo can run your tests with the cargo test command. Cargo looks for tests to run in two places: in each of your src files and any tests in tests/. Tests in your src files should be unit tests, and tests in tests/ should be integration-style tests. As such, you’ll need to import your crates into the files in tests.

Here's an example of running cargo test in our package, which currently has no tests:

$ cargo test
   Compiling rand v0.1.0 (https://github.com/rust-lang-nursery/rand.git#9f35b8e)
   Compiling hello_world v0.1.0 (file:///path/to/package/hello_world)
     Running target/test/hello_world-9c2b65bbb79eabce

running 0 tests

test result: ok. 0 passed; 0 failed; 0 ignored; 0 measured; 0 filtered out

If our package had tests, we would see more output with the correct number of tests.

You can also run a specific test by passing a filter:

$ cargo test foo

This will run any test with foo in its name.

cargo test runs additional checks as well. For example, it will compile any examples you’ve included and will also test the examples in your documentation. Please see the testing guide in the Rust documentation for more details.

Continuous Integration

Travis CI

To test your package on Travis CI, here is a sample .travis.yml file:

language: rust
rust:
  - stable
  - beta
  - nightly
matrix:
  allow_failures:
    - rust: nightly

This will test all three release channels, but any breakage in nightly will not fail your overall build. Please see the Travis CI Rust documentation for more information.

GitLab CI

To test your package on GitLab CI, here is a sample .gitlab-ci.yml file:

stages:
  - build

rust-latest:
  stage: build
  image: rust:latest
  script:
    - cargo build --verbose
    - cargo test --verbose

rust-nightly:
  stage: build
  image: rustlang/rust:nightly
  script:
    - cargo build --verbose
    - cargo test --verbose
  allow_failure: true

This will test on the stable channel and nightly channel, but any breakage in nightly will not fail your overall build. Please see the GitLab CI for more information.

builds.sr.ht

To test your package on sr.ht, here is a sample .build.yml file. Be sure to change <your repo> and <your project> to the repo to clone and the directory where it was cloned.

image: archlinux
packages:
  - rustup
sources:
  - <your repo>
tasks:
  - setup: |
      rustup toolchain install nightly stable
      cd <your project>/
      rustup run stable cargo fetch
  - stable: |
      rustup default stable
      cd <your project>/
      cargo build --verbose
      cargo test --verbose
  - nightly: |
      rustup default nightly
      cd <your project>/
      cargo build --verbose ||:
      cargo test --verbose  ||:
  - docs: |
      cd <your project>/
      rustup run stable cargo doc --no-deps
      rustup run nightly cargo doc --no-deps ||:

This will test and build documentation on the stable channel and nightly channel, but any breakage in nightly will not fail your overall build. Please see the builds.sr.ht documentation for more information.

Cargo Home

The "Cargo home" functions as a download and source cache. When building a crate, Cargo stores downloaded build dependencies in the Cargo home. You can alter the location of the Cargo home by setting the CARGO_HOME environmental variable. The home crate provides an API for getting this location if you need this information inside your Rust crate. By default, the Cargo home is located in $HOME/.cargo/.

Please note that the internal structure of the Cargo home is not stabilized and may be subject to change at any time.

The Cargo home consists of following components:

Files:

• config Cargo's global configuration file, see the config entry in the reference.

• credentials Private login credentials from cargo login in order to log in to a registry.

• .crates.toml This hidden file contains package information of crates installed via cargo install. Do NOT edit by hand!

Directories:

• bin The bin directory contains executables of crates that were installed via cargo install or rustup. To be able to make these binaries accessible, add the path of the directory to your $PATH environment variable.

• git Git sources are stored here:

  • git/db When a crate depends on a git repository, Cargo clones the repo as a bare repo into this directory and updates it if necessary.

  • git/checkouts If a git source is used, the required commit of the repo is checked out from the bare repo inside git/db into this directory. This provides the compiler with the actual files contained in the repo of the commit specified for that dependency. Multiple checkouts of different commits of the same repo are possible.

• registry Packages and metadata of crate registries (such as crates.io) are located here.

  • registry/index The index is a bare git repository which contains the metadata (versions, dependencies etc) of all available crates of a registry.

  • registry/cache Downloaded dependencies are stored in the cache. The crates are compressed gzip archives named with a .crate extension.

  • registry/src If a downloaded .crate archive is required by a package, it is unpacked into registry/src folder where rustc will find the .rs files.

Caching the Cargo home in CI

To avoid redownloading all crate dependencies during continuous integration, you can cache the $CARGO_HOME directory. However, caching the entire directory is often inefficient as it will contain downloaded sources twice. If we depend on a crate such as serde 1.0.92 and cache the entire $CARGO_HOME we would actually cache the sources twice, the serde-1.0.92.crate inside registry/cache and the extracted .rs files of serde inside registry/src. The can unnecessarily slow down the build as downloading, extracting, recompressing and reuploading the cache to the CI servers can take some time.

It should be sufficient to only cache the following directories across builds:

• bin/
• registry/index/
• registry/cache/
• git/db/

Vendoring all dependencies of a project

See the cargo vendor subcommand.

Clearing the cache

In theory, you can always remove any part of the cache and Cargo will do its best to restore sources if a crate needs them either by reextracting an archive or checking out a bare repo or by simply redownloading the sources from the web.

Alternatively, the cargo-cache crate provides a simple CLI tool to only clear selected parts of the cache or show sizes of its components in your command-line.

Build cache

Cargo shares build artifacts among all the packages of a single workspace. Today, Cargo does not share build results across different workspaces, but a similar result can be achieved by using a third party tool, sccache.

To setup sccache, install it with cargo install sccache and set RUSTC_WRAPPER environmental variable to sccache before invoking Cargo. If you use bash, it makes sense to add export RUSTC_WRAPPER=sccache to .bashrc. Refer to sccache documentation for more details.

Cargo Reference

The reference covers the details of various areas of Cargo.

• Specifying Dependencies
• The Manifest Format
• Configuration
• Environment Variables
• Build Scripts
• Publishing on crates.io
• Package ID Specifications
• Source Replacement
• External Tools
• Unstable Features

Specifying Dependencies

Your crates can depend on other libraries from crates.io or other registries, git repositories, or subdirectories on your local file system. You can also temporarily override the location of a dependency — for example, to be able to test out a bug fix in the dependency that you are working on locally. You can have different dependencies for different platforms, and dependencies that are only used during development. Let's take a look at how to do each of these.

Specifying dependencies from crates.io

Cargo is configured to look for dependencies on crates.io by default. Only the name and a version string are required in this case. In the cargo guide, we specified a dependency on the time crate:

[dependencies]
time = "0.1.12"

The string "0.1.12" is a semver version requirement. Since this string does not have any operators in it, it is interpreted the same way as if we had specified "^0.1.12", which is called a caret requirement.

Caret requirements

Caret requirements allow SemVer compatible updates to a specified version. An update is allowed if the new version number does not modify the left-most non-zero digit in the major, minor, patch grouping. In this case, if we ran cargo update -p time, cargo should update us to version 0.1.13 if it is the latest 0.1.z release, but would not update us to 0.2.0. If instead we had specified the version string as ^1.0, cargo should update to 1.1 if it is the latest 1.y release, but not 2.0. The version 0.0.x is not considered compatible with any other version.

Here are some more examples of caret requirements and the versions that would be allowed with them:

^1.2.3  :=  >=1.2.3 <2.0.0
^1.2    :=  >=1.2.0 <2.0.0
^1      :=  >=1.0.0 <2.0.0
^0.2.3  :=  >=0.2.3 <0.3.0
^0.2    :=  >=0.2.0 <0.3.0
^0.0.3  :=  >=0.0.3 <0.0.4
^0.0    :=  >=0.0.0 <0.1.0
^0      :=  >=0.0.0 <1.0.0

This compatibility convention is different from SemVer in the way it treats versions before 1.0.0. While SemVer says there is no compatibility before 1.0.0, Cargo considers 0.x.y to be compatible with 0.x.z, where y ≥ z and x > 0.

Tilde requirements

Tilde requirements specify a minimal version with some ability to update. If you specify a major, minor, and patch version or only a major and minor version, only patch-level changes are allowed. If you only specify a major version, then minor- and patch-level changes are allowed.

~1.2.3 is an example of a tilde requirement.

~1.2.3  := >=1.2.3 <1.3.0
~1.2    := >=1.2.0 <1.3.0
~1      := >=1.0.0 <2.0.0

Wildcard requirements

Wildcard requirements allow for any version where the wildcard is positioned.

*, 1.* and 1.2.* are examples of wildcard requirements.

*     := >=0.0.0
1.*   := >=1.0.0 <2.0.0
1.2.* := >=1.2.0 <1.3.0

Comparison requirements

Comparison requirements allow manually specifying a version range or an exact version to depend on.

Here are some examples of comparison requirements:

>= 1.2.0
> 1
< 2
= 1.2.3

Multiple requirements

Multiple version requirements can also be separated with a comma, e.g., >= 1.2, < 1.5.

Specifying dependencies from other registries

To specify a dependency from a registry other than crates.io, first the registry must be configured in a .cargo/config file. See the registries documentation for more information. In the dependency, set the registry key to the name of the registry to use.

[dependencies]
some-crate = { version = "1.0", registry = "my-registry" }

Specifying dependencies from git repositories

To depend on a library located in a git repository, the minimum information you need to specify is the location of the repository with the git key:

[dependencies]
rand = { git = "https://github.com/rust-lang-nursery/rand" }

Cargo will fetch the git repository at this location then look for a Cargo.toml for the requested crate anywhere inside the git repository (not necessarily at the root - for example, specifying a member crate name of a workspace and setting git to the repository containing the workspace).

Since we haven’t specified any other information, Cargo assumes that we intend to use the latest commit on the master branch to build our package. You can combine the git key with the rev, tag, or branch keys to specify something else. Here's an example of specifying that you want to use the latest commit on a branch named next:

[dependencies]
rand = { git = "https://github.com/rust-lang-nursery/rand", branch = "next" }

Specifying path dependencies

Over time, our hello_world package from the guide has grown significantly in size! It’s gotten to the point that we probably want to split out a separate crate for others to use. To do this Cargo supports path dependencies which are typically sub-crates that live within one repository. Let’s start off by making a new crate inside of our hello_world package:

# inside of hello_world/
$ cargo new hello_utils

This will create a new folder hello_utils inside of which a Cargo.toml and src folder are ready to be configured. In order to tell Cargo about this, open up hello_world/Cargo.toml and add hello_utils to your dependencies:

[dependencies]
hello_utils = { path = "hello_utils" }

This tells Cargo that we depend on a crate called hello_utils which is found in the hello_utils folder (relative to the Cargo.toml it’s written in).

And that’s it! The next cargo build will automatically build hello_utils and all of its own dependencies, and others can also start using the crate as well. However, crates that use dependencies specified with only a path are not permitted on crates.io. If we wanted to publish our hello_world crate, we would need to publish a version of hello_utils to crates.io and specify its version in the dependencies line as well:

[dependencies]
hello_utils = { path = "hello_utils", version = "0.1.0" }

Overriding dependencies

There are a number of methods in Cargo to support overriding dependencies and otherwise controlling the dependency graph. These options are typically, though, only available at the workspace level and aren't propagated through dependencies. In other words, "applications" have the ability to override dependencies but "libraries" do not.

The desire to override a dependency or otherwise alter some dependencies can arise through a number of scenarios. Most of them, however, boil down to the ability to work with a crate before it's been published to crates.io. For example:

• A crate you're working on is also used in a much larger application you're working on, and you'd like to test a bug fix to the library inside of the larger application.
• An upstream crate you don't work on has a new feature or a bug fix on the master branch of its git repository which you'd like to test out.
• You're about to publish a new major version of your crate, but you'd like to do integration testing across an entire package to ensure the new major version works.
• You've submitted a fix to an upstream crate for a bug you found, but you'd like to immediately have your application start depending on the fixed version of the crate to avoid blocking on the bug fix getting merged.

These scenarios are currently all solved with the [patch] manifest section. Historically some of these scenarios have been solved with the [replace] section, but we'll document the [patch] section here.

Testing a bugfix

Let's say you're working with the uuid crate but while you're working on it you discover a bug. You are, however, quite enterprising so you decide to also try to fix the bug! Originally your manifest will look like:

[package]
name = "my-library"
version = "0.1.0"
authors = ["..."]

[dependencies]
uuid = "1.0"

First thing we'll do is to clone the uuid repository locally via:

$ git clone https://github.com/rust-lang-nursery/uuid

Next we'll edit the manifest of my-library to contain:

[patch.crates-io]
uuid = { path = "../path/to/uuid" }

Here we declare that we're patching the source crates-io with a new dependency. This will effectively add the local checked out version of uuid to the crates.io registry for our local package.

Next up we need to ensure that our lock file is updated to use this new version of uuid so our package uses the locally checked out copy instead of one from crates.io. The way [patch] works is that it'll load the dependency at ../path/to/uuid and then whenever crates.io is queried for versions of uuid it'll also return the local version.

This means that the version number of the local checkout is significant and will affect whether the patch is used. Our manifest declared uuid = "1.0" which means we'll only resolve to >= 1.0.0, < 2.0.0, and Cargo's greedy resolution algorithm also means that we'll resolve to the maximum version within that range. Typically this doesn't matter as the version of the git repository will already be greater or match the maximum version published on crates.io, but it's important to keep this in mind!

In any case, typically all you need to do now is:

$ cargo build
   Compiling uuid v1.0.0 (.../uuid)
   Compiling my-library v0.1.0 (.../my-library)
    Finished dev [unoptimized + debuginfo] target(s) in 0.32 secs

And that's it! You're now building with the local version of uuid (note the path in parentheses in the build output). If you don't see the local path version getting built then you may need to run cargo update -p uuid --precise $version where $version is the version of the locally checked out copy of uuid.

Once you've fixed the bug you originally found the next thing you'll want to do is to likely submit that as a pull request to the uuid crate itself. Once you've done this then you can also update the [patch] section. The listing inside of [patch] is just like the [dependencies] section, so once your pull request is merged you could change your path dependency to:

[patch.crates-io]
uuid = { git = 'https://github.com/rust-lang-nursery/uuid' }

Working with an unpublished minor version

Let's now shift gears a bit from bug fixes to adding features. While working on my-library you discover that a whole new feature is needed in the uuid crate. You've implemented this feature, tested it locally above with [patch], and submitted a pull request. Let's go over how you continue to use and test it before it's actually published.

Let's also say that the current version of uuid on crates.io is 1.0.0, but since then the master branch of the git repository has updated to 1.0.1. This branch includes your new feature you submitted previously. To use this repository we'll edit our Cargo.toml to look like

[package]
name = "my-library"
version = "0.1.0"
authors = ["..."]

[dependencies]
uuid = "1.0.1"

[patch.crates-io]
uuid = { git = 'https://github.com/rust-lang-nursery/uuid' }

Note that our local dependency on uuid has been updated to 1.0.1 as it's what we'll actually require once the crate is published. This version doesn't exist on crates.io, though, so we provide it with the [patch] section of the manifest.

Now when our library is built it'll fetch uuid from the git repository and resolve to 1.0.1 inside the repository instead of trying to download a version from crates.io. Once 1.0.1 is published on crates.io the [patch] section can be deleted.

It's also worth noting that [patch] applies transitively. Let's say you use my-library in a larger package, such as:

[package]
name = "my-binary"
version = "0.1.0"
authors = ["..."]

[dependencies]
my-library = { git = 'https://example.com/git/my-library' }
uuid = "1.0"

[patch.crates-io]
uuid = { git = 'https://github.com/rust-lang-nursery/uuid' }

Remember that [patch] is applicable transitively but can only be defined at the top level so we consumers of my-library have to repeat the [patch] section if necessary. Here, though, the new uuid crate applies to both our dependency on uuid and the my-library -> uuid dependency. The uuid crate will be resolved to one version for this entire crate graph, 1.0.1, and it'll be pulled from the git repository.

Overriding repository URL

In case the dependency you want to override isn't loaded from crates.io, you'll have to change a bit how you use [patch]:

[patch."https://github.com/your/repository"]
my-library = { path = "../my-library/path" }

And that's it!

Prepublishing a breaking change

As a final scenario, let's take a look at working with a new major version of a crate, typically accompanied with breaking changes. Sticking with our previous crates, this means that we're going to be creating version 2.0.0 of the uuid crate. After we've submitted all changes upstream we can update our manifest for my-library to look like:

[dependencies]
uuid = "2.0"

[patch.crates-io]
uuid = { git = "https://github.com/rust-lang-nursery/uuid", branch = "2.0.0" }

And that's it! Like with the previous example the 2.0.0 version doesn't actually exist on crates.io but we can still put it in through a git dependency through the usage of the [patch] section. As a thought exercise let's take another look at the my-binary manifest from above again as well:

[package]
name = "my-binary"
version = "0.1.0"
authors = ["..."]

[dependencies]
my-library = { git = 'https://example.com/git/my-library' }
uuid = "1.0"

[patch.crates-io]
uuid = { git = 'https://github.com/rust-lang-nursery/uuid', branch = '2.0.0' }

Note that this will actually resolve to two versions of the uuid crate. The my-binary crate will continue to use the 1.x.y series of the uuid crate but the my-library crate will use the 2.0.0 version of uuid. This will allow you to gradually roll out breaking changes to a crate through a dependency graph without being force to update everything all at once.

Overriding with local dependencies

Sometimes you're only temporarily working on a crate and you don't want to have to modify Cargo.toml like with the [patch] section above. For this use case Cargo offers a much more limited version of overrides called path overrides.

Path overrides are specified through .cargo/config instead of Cargo.toml, and you can find more documentation about this configuration. Inside of .cargo/config you'll specify a key called paths:

paths = ["/path/to/uuid"]

This array should be filled with directories that contain a Cargo.toml. In this instance, we’re just adding uuid, so it will be the only one that’s overridden. This path can be either absolute or relative to the directory that contains the .cargo folder.

Path overrides are more restricted than the [patch] section, however, in that they cannot change the structure of the dependency graph. When a path replacement is used then the previous set of dependencies must all match exactly to the new Cargo.toml specification. For example this means that path overrides cannot be used to test out adding a dependency to a crate, instead [patch] must be used in that situation. As a result usage of a path override is typically isolated to quick bug fixes rather than larger changes.

Note: using a local configuration to override paths will only work for crates that have been published to crates.io. You cannot use this feature to tell Cargo how to find local unpublished crates.

Platform specific dependencies

Platform-specific dependencies take the same format, but are listed under a target section. Normally Rust-like #[cfg] syntax will be used to define these sections:

[target.'cfg(windows)'.dependencies]
winhttp = "0.4.0"

[target.'cfg(unix)'.dependencies]
openssl = "1.0.1"

[target.'cfg(target_arch = "x86")'.dependencies]
native = { path = "native/i686" }

[target.'cfg(target_arch = "x86_64")'.dependencies]
native = { path = "native/x86_64" }

Like with Rust, the syntax here supports the not, any, and all operators to combine various cfg name/value pairs.

If you want to know which cfg targets are available on your platform, run rustc --print=cfg from the command line. If you want to know which cfg targets are available for another platform, such as 64-bit Windows, run rustc --print=cfg --target=x86_64-pc-windows-msvc.

Unlike in your Rust source code, you cannot use [target.'cfg(feature = "my_crate")'.dependencies] to add dependencies based on optional crate features. Use the [features] section instead.

In addition to #[cfg] syntax, Cargo also supports listing out the full target the dependencies would apply to:

[target.x86_64-pc-windows-gnu.dependencies]
winhttp = "0.4.0"

[target.i686-unknown-linux-gnu.dependencies]
openssl = "1.0.1"

If you’re using a custom target specification, quote the full path and file name:

[target."x86_64/windows.json".dependencies]
winhttp = "0.4.0"

[target."i686/linux.json".dependencies]
openssl = "1.0.1"
native = { path = "native/i686" }

[target."x86_64/linux.json".dependencies]
openssl = "1.0.1"
native = { path = "native/x86_64" }

Development dependencies

You can add a [dev-dependencies] section to your Cargo.toml whose format is equivalent to [dependencies]:

[dev-dependencies]
tempdir = "0.3"

Dev-dependencies are not used when compiling a package for building, but are used for compiling tests, examples, and benchmarks.

These dependencies are not propagated to other packages which depend on this package.

You can also have target-specific development dependencies by using dev-dependencies in the target section header instead of dependencies. For example:

[target.'cfg(unix)'.dev-dependencies]
mio = "0.0.1"

Build dependencies

You can depend on other Cargo-based crates for use in your build scripts. Dependencies are declared through the build-dependencies section of the manifest:

[build-dependencies]
cc = "1.0.3"

The build script does not have access to the dependencies listed in the dependencies or dev-dependencies section. Build dependencies will likewise not be available to the package itself unless listed under the dependencies section as well. A package itself and its build script are built separately, so their dependencies need not coincide. Cargo is kept simpler and cleaner by using independent dependencies for independent purposes.

Choosing features

If a package you depend on offers conditional features, you can specify which to use:

[dependencies.awesome]
version = "1.3.5"
default-features = false # do not include the default features, and optionally
                         # cherry-pick individual features
features = ["secure-password", "civet"]

More information about features can be found in the manifest documentation.

Renaming dependencies in Cargo.toml

When writing a [dependencies] section in Cargo.toml the key you write for a dependency typically matches up to the name of the crate you import from in the code. For some projects, though, you may wish to reference the crate with a different name in the code regardless of how it's published on crates.io. For example you may wish to:

• Avoid the need to use foo as bar in Rust source.
• Depend on multiple versions of a crate.
• Depend on crates with the same name from different registries.

To support this Cargo supports a package key in the [dependencies] section of which package should be depended on:

[package]
name = "mypackage"
version = "0.0.1"

[dependencies]
foo = "0.1"
bar = { git = "https://github.com/example/project", package = "foo" }
baz = { version = "0.1", registry = "custom", package = "foo" }

In this example, three crates are now available in your Rust code:

# #![allow(unused_variables)]
#fn main() {
extern crate foo; // crates.io
extern crate bar; // git repository
extern crate baz; // registry `custom`
#}

All three of these crates have the package name of foo in their own Cargo.toml, so we're explicitly using the package key to inform Cargo that we want the foo package even though we're calling it something else locally. The package key, if not specified, defaults to the name of the dependency being requested.

Note that if you have an optional dependency like:

[dependencies]
foo = { version = "0.1", package = 'bar', optional = true }

you're depending on the crate bar from crates.io, but your crate has a foo feature instead of a bar feature. That is, names of features take after the name of the dependency, not the package name, when renamed.

Enabling transitive dependencies works similarly, for example we could add the following to the above manifest:

[features]
log-debug = ['foo/log-debug'] # using 'bar/log-debug' would be an error!

The Manifest Format

The Cargo.toml file for each package is called its manifest. Every manifest file consists of one or more sections.

The [package] section

The first section in a Cargo.toml is [package].

[package]
name = "hello_world" # the name of the package
version = "0.1.0"    # the current version, obeying semver
authors = ["Alice <a@example.com>", "Bob <b@example.com>"]

The name field

The package name is an identifier used to refer to the package. It is used when listed as a dependency in another package, and as the default name of inferred lib and bin targets.

The name must not be empty, use only alphanumeric characters or - or _. Note that cargo new and cargo init impose some additional restrictions on the package name, such as enforcing that it is a valid Rust identifier and not a keyword. crates.io imposes even more restrictions, such as enforcing only ASCII characters, not a reserved name, not a special Windows name such as "nul", is not too long, etc.

The version field

Cargo bakes in the concept of Semantic Versioning, so make sure you follow some basic rules:

• Before you reach 1.0.0, anything goes, but if you make breaking changes, increment the minor version. In Rust, breaking changes include adding fields to structs or variants to enums.
• After 1.0.0, only make breaking changes when you increment the major version. Don’t break the build.
• After 1.0.0, don’t add any new public API (no new pub anything) in patch-level versions. Always increment the minor version if you add any new pub structs, traits, fields, types, functions, methods or anything else.
• Use version numbers with three numeric parts such as 1.0.0 rather than 1.0.

The authors field (optional)

The authors field lists people or organizations that are considered the "authors" of the package. The exact meaning is open to interpretation — it may list the original or primary authors, current maintainers, or owners of the package. These names will be listed on the crate's page on crates.io. An optional email address may be included within angled brackets at the end of each author.

The edition field (optional)

You can opt in to a specific Rust Edition for your package with the edition key in Cargo.toml. If you don't specify the edition, it will default to 2015.

[package]
# ...
edition = '2018'

The edition key affects which edition your package is compiled with. Cargo will always generate packages via cargo new with the edition key set to the latest edition. Setting the edition key in [package] will affect all targets/crates in the package, including test suites, benchmarks, binaries, examples, etc.

The build field (optional)

This field specifies a file in the package root which is a build script for building native code. More information can be found in the build script guide.

[package]
# ...
build = "build.rs"

The links field (optional)

This field specifies the name of a native library that is being linked to. More information can be found in the links section of the build script guide.

[package]
# ...
links = "foo"
build = "build.rs"

The documentation field (optional)

This field specifies a URL to a website hosting the crate's documentation. If no URL is specified in the manifest file, crates.io will automatically link your crate to the corresponding docs.rs page.

Documentation links from specific hosts are blacklisted. Hosts are added to the blacklist if they are known to not be hosting documentation and are possibly of malicious intent e.g., ad tracking networks. URLs from the following hosts are blacklisted:

• rust-ci.org

Documentation URLs from blacklisted hosts will not appear on crates.io, and may be replaced by docs.rs links.

The exclude and include fields (optional)

You can explicitly specify that a set of file patterns should be ignored or included for the purposes of packaging. The patterns specified in the exclude field identify a set of files that are not included, and the patterns in include specify files that are explicitly included.

The patterns should be gitignore-style patterns. Briefly:

• foo matches any file or directory with the name foo anywhere in the package. This is equivalent to the pattern **/foo.
• /foo matches any file or directory with the name foo only in the root of the package.
• foo/ matches any directory with the name foo anywhere in the package.
• Common glob patterns like *, ?, and [] are supported:
  • * matches zero or more characters except /. For example, *.html matches any file or directory with the .html extension anywhere in the package.
  • ? matches any character except /. For example, foo? matches food, but not foo.
  • [] allows for matching a range of characters. For example, [ab] matches either a or b. [a-z] matches letters a through z.
• **/ prefix matches in any directory. For example, **/foo/bar matches the file or directory bar anywhere that is directly under directory foo.
• /** suffix matches everything inside. For example, foo/** matches all files inside directory foo, including all files in subdirectories below foo.
• /**/ matches zero or more directories. For example, a/**/b matches a/b, a/x/b, a/x/y/b, and so on.
• ! prefix negates a pattern. For example, a pattern of src/**.rs and !foo.rs would match all files with the .rs extension inside the src directory, except for any file named foo.rs.

If git is being used for a package, the exclude field will be seeded with the gitignore settings from the repository.

[package]
# ...
exclude = ["build/**/*.o", "doc/**/*.html"]

[package]
# ...
include = ["src/**/*", "Cargo.toml"]

The options are mutually exclusive: setting include will override an exclude. Note that include must be an exhaustive list of files as otherwise necessary source files may not be included. The package's Cargo.toml is automatically included.

The include/exclude list is also used for change tracking in some situations. For targets built with rustdoc, it is used to determine the list of files to track to determine if the target should be rebuilt. If the package has a build script that does not emit any rerun-if-* directives, then the include/exclude list is used for tracking if the build script should be re-run if any of those files change.

The publish field (optional)

The publish field can be used to prevent a package from being published to a package registry (like crates.io) by mistake, for instance to keep a package private in a company.

[package]
# ...
publish = false

The value may also be an array of strings which are registry names that are allowed to be published to.

[package]
# ...
publish = ["some-registry-name"]

The workspace field (optional)

The workspace field can be used to configure the workspace that this package will be a member of. If not specified this will be inferred as the first Cargo.toml with [workspace] upwards in the filesystem.

[package]
# ...
workspace = "path/to/workspace/root"

For more information, see the documentation for the workspace table below.

Package metadata

There are a number of optional metadata fields also accepted under the [package] section:

[package]
# ...

# A short blurb about the package. This is not rendered in any format when
# uploaded to crates.io (aka this is not markdown).
description = "..."

# These URLs point to more information about the package. These are
# intended to be webviews of the relevant data, not necessarily compatible
# with VCS tools and the like.
documentation = "..."
homepage = "..."
repository = "..."

# This points to a file under the package root (relative to this `Cargo.toml`).
# The contents of this file are stored and indexed in the registry.
# crates.io will render this file and place the result on the crate's page.
readme = "..."

# This is a list of up to five keywords that describe this crate. Keywords
# are searchable on crates.io, and you may choose any words that would
# help someone find this crate.
keywords = ["...", "..."]

# This is a list of up to five categories where this crate would fit.
# Categories are a fixed list available at crates.io/category_slugs, and
# they must match exactly.
categories = ["...", "..."]

# This is an SPDX 2.1 license expression for this package. Currently
# crates.io will validate the license provided against a whitelist of
# known license and exception identifiers from the SPDX license list
# 2.4. Parentheses are not currently supported.
#
# Multiple licenses can be separated with a `/`, although that usage
# is deprecated. Instead, use a license expression with AND and OR
# operators to get more explicit semantics.
license = "..."

# If a package is using a nonstandard license, then this key may be specified in
# lieu of the above key and must point to a file relative to this manifest
# (similar to the readme key).
license-file = "..."

# Optional specification of badges to be displayed on crates.io.
#
# - The badges pertaining to build status that are currently available are
#   Appveyor, CircleCI, Cirrus CI, GitLab, Azure DevOps and TravisCI.
# - Available badges pertaining to code test coverage are Codecov and
#   Coveralls.
# - There are also maintenance-related badges based on isitmaintained.com
#   which state the issue resolution time, percent of open issues, and future
#   maintenance intentions.
#
# If a `repository` key is required, this refers to a repository in
# `user/repo` format.
[badges]

# Appveyor: `repository` is required. `branch` is optional; default is `master`
# `service` is optional; valid values are `github` (default), `bitbucket`, and
# `gitlab`; `id` is optional; you can specify the appveyor project id if you
# want to use that instead. `project_name` is optional; use when the repository
# name differs from the appveyor project name.
appveyor = { repository = "...", branch = "master", service = "github" }

# Circle CI: `repository` is required. `branch` is optional; default is `master`
circle-ci = { repository = "...", branch = "master" }

# Cirrus CI: `repository` is required. `branch` is optional; default is `master`
cirrus-ci = { repository = "...", branch = "master" }

# GitLab: `repository` is required. `branch` is optional; default is `master`
gitlab = { repository = "...", branch = "master" }

# Azure DevOps: `project` is required. `pipeline` is required. `build` is optional; default is `1`
# Note: project = `organization/project`, pipeline = `name_of_pipeline`, build = `definitionId`
azure-devops = { project = "...", pipeline = "...", build="2" }

# Travis CI: `repository` in format "<user>/<project>" is required.
# `branch` is optional; default is `master`
travis-ci = { repository = "...", branch = "master" }

# Codecov: `repository` is required. `branch` is optional; default is `master`
# `service` is optional; valid values are `github` (default), `bitbucket`, and
# `gitlab`.
codecov = { repository = "...", branch = "master", service = "github" }

# Coveralls: `repository` is required. `branch` is optional; default is `master`
# `service` is optional; valid values are `github` (default) and `bitbucket`.
coveralls = { repository = "...", branch = "master", service = "github" }

# Is it maintained resolution time: `repository` is required.
is-it-maintained-issue-resolution = { repository = "..." }

# Is it maintained percentage of open issues: `repository` is required.
is-it-maintained-open-issues = { repository = "..." }

# Maintenance: `status` is required. Available options are:
# - `actively-developed`: New features are being added and bugs are being fixed.
# - `passively-maintained`: There are no plans for new features, but the maintainer intends to
#   respond to issues that get filed.
# - `as-is`: The crate is feature complete, the maintainer does not intend to continue working on
#   it or providing support, but it works for the purposes it was designed for.
# - `experimental`: The author wants to share it with the community but is not intending to meet
#   anyone's particular use case.
# - `looking-for-maintainer`: The current maintainer would like to transfer the crate to someone
#   else.
# - `deprecated`: The maintainer does not recommend using this crate (the description of the crate
#   can describe why, there could be a better solution available or there could be problems with
#   the crate that the author does not want to fix).
# - `none`: Displays no badge on crates.io, since the maintainer has not chosen to specify
#   their intentions, potential crate users will need to investigate on their own.
maintenance = { status = "..." }

The crates.io registry will render the description, display the license, link to the three URLs and categorize by the keywords. These keys provide useful information to users of the registry and also influence the search ranking of a crate. It is highly discouraged to omit everything in a published crate.

SPDX 2.1 license expressions are documented here. The current version of the license list is available here, and version 2.4 is available here.

The metadata table (optional)

Cargo by default will warn about unused keys in Cargo.toml to assist in detecting typos and such. The package.metadata table, however, is completely ignored by Cargo and will not be warned about. This section can be used for tools which would like to store package configuration in Cargo.toml. For example:

[package]
name = "..."
# ...

# Metadata used when generating an Android APK, for example.
[package.metadata.android]
package-name = "my-awesome-android-app"
assets = "path/to/static"

The default-run field

The default-run field in the [package] section of the manifest can be used to specify a default binary picked by cargo run. For example, when there is both src/bin/a.rs and src/bin/b.rs:

[package]
default-run = "a"

Dependency sections

See the specifying dependencies page for information on the [dependencies], [dev-dependencies], [build-dependencies], and target-specific [target.*.dependencies] sections.

The [profile.*] sections

Cargo supports custom configuration of how rustc is invoked through profiles at the top level. Any manifest may declare a profile, but only the top level package’s profiles are actually read. All dependencies’ profiles will be overridden. This is done so the top-level package has control over how its dependencies are compiled.

There are four currently supported profile names, all of which have the same configuration available to them. Listed below is the configuration available, along with the defaults for each profile.

# The development profile, used for `cargo build`.
[profile.dev]
opt-level = 0      # controls the `--opt-level` the compiler builds with.
                   # 0-1 is good for debugging. 2 is well-optimized. Max is 3.
                   # 's' attempts to reduce size, 'z' reduces size even more.
debug = true       # (u32 or bool) Include debug information (debug symbols).
                   # Equivalent to `-C debuginfo=2` compiler flag.
rpath = false      # controls whether compiler should set loader paths.
                   # If true, passes `-C rpath` flag to the compiler.
lto = false        # Link Time Optimization usually reduces size of binaries
                   # and static libraries. Increases compilation time.
                   # If true, passes `-C lto` flag to the compiler, and if a
                   # string is specified like 'thin' then `-C lto=thin` will
                   # be passed.
debug-assertions = true # controls whether debug assertions are enabled
                   # (e.g., debug_assert!() and arithmetic overflow checks)
codegen-units = 16 # if > 1 enables parallel code generation which improves
                   # compile times, but prevents some optimizations.
                   # Passes `-C codegen-units`.
panic = 'unwind'   # panic strategy (`-C panic=...`), can also be 'abort'
incremental = true # whether or not incremental compilation is enabled
                   # This can be overridden globally with the CARGO_INCREMENTAL
                   # environment variable or `build.incremental` config
                   # variable. Incremental is only used for path sources.
overflow-checks = true # use overflow checks for integer arithmetic.
                   # Passes the `-C overflow-checks=...` flag to the compiler.

# The release profile, used for `cargo build --release` (and the dependencies
# for `cargo test --release`, including the local library or binary).
[profile.release]
opt-level = 3
debug = false
rpath = false
lto = false
debug-assertions = false
codegen-units = 16
panic = 'unwind'
incremental = false
overflow-checks = false

# The testing profile, used for `cargo test` (for `cargo test --release` see
# the `release` and `bench` profiles).
[profile.test]
opt-level = 0
debug = 2
rpath = false
lto = false
debug-assertions = true
codegen-units = 16
panic = 'unwind'
incremental = true
overflow-checks = true

# The benchmarking profile, used for `cargo bench` (and the test targets and
# unit tests for `cargo test --release`).
[profile.bench]
opt-level = 3
debug = false
rpath = false
lto = false
debug-assertions = false
codegen-units = 16
panic = 'unwind'
incremental = false
overflow-checks = false

The [features] section

Cargo supports features to allow expression of:

• conditional compilation options (usable through cfg attributes);
• optional dependencies, which enhance a package, but are not required; and
• clusters of optional dependencies, such as postgres, that would include the postgres package, the postgres-macros package, and possibly other packages (such as development-time mocking libraries, debugging tools, etc.).

A feature of a package is either an optional dependency, or a set of other features. The format for specifying features is:

[package]
name = "awesome"

[features]
# The default set of optional packages. Most people will want to use these
# packages, but they are strictly optional. Note that `session` is not a package
# but rather another feature listed in this manifest.
default = ["jquery", "uglifier", "session"]

# A feature with no dependencies is used mainly for conditional compilation,
# like `#[cfg(feature = "go-faster")]`.
go-faster = []

# The `secure-password` feature depends on the bcrypt package. This aliasing
# will allow people to talk about the feature in a higher-level way and allow
# this package to add more requirements to the feature in the future.
secure-password = ["bcrypt"]

# Features can be used to reexport features of other packages. The `session`
# feature of package `awesome` will ensure that the `session` feature of the
# package `cookie` is also enabled.
session = ["cookie/session"]

[dependencies]
# These packages are mandatory and form the core of this package’s distribution.
cookie = "1.2.0"
oauth = "1.1.0"
route-recognizer = "=2.1.0"

# A list of all of the optional dependencies, some of which are included in the
# above `features`. They can be opted into by apps.
jquery = { version = "1.0.2", optional = true }
uglifier = { version = "1.5.3", optional = true }
bcrypt = { version = "*", optional = true }
civet = { version = "*", optional = true }

To use the package awesome:

[dependencies.awesome]
version = "1.3.5"
default-features = false # do not include the default features, and optionally
                         # cherry-pick individual features
features = ["secure-password", "civet"]

Rules

The usage of features is subject to a few rules:

• Feature names must not conflict with other package names in the manifest. This is because they are opted into via features = [...], which only has a single namespace.
• With the exception of the default feature, all features are opt-in. To opt out of the default feature, use default-features = false and cherry-pick individual features.
• Feature groups are not allowed to cyclically depend on one another.
• Dev-dependencies cannot be optional.
• Features groups can only reference optional dependencies.
• When a feature is selected, Cargo will call rustc with --cfg feature="${feature_name}". If a feature group is included, it and all of its individual features will be included. This can be tested in code via #[cfg(feature = "foo")].

Note that it is explicitly allowed for features to not actually activate any optional dependencies. This allows packages to internally enable/disable features without requiring a new dependency.

Usage in end products

One major use-case for this feature is specifying optional features in end-products. For example, the Servo package may want to include optional features that people can enable or disable when they build it.

In that case, Servo will describe features in its Cargo.toml and they can be enabled using command-line flags:

$ cargo build --release --features "shumway pdf"

Default features could be excluded using --no-default-features.

Usage in packages

In most cases, the concept of optional dependency in a library is best expressed as a separate package that the top-level application depends on.

However, high-level packages, like Iron or Piston, may want the ability to curate a number of packages for easy installation. The current Cargo system allows them to curate a number of mandatory dependencies into a single package for easy installation.

In some cases, packages may want to provide additional curation for optional dependencies:

• grouping a number of low-level optional dependencies together into a single high-level feature;
• specifying packages that are recommended (or suggested) to be included by users of the package; and
• including a feature (like secure-password in the motivating example) that will only work if an optional dependency is available, and would be difficult to implement as a separate package (for example, it may be overly difficult to design an IO package to be completely decoupled from OpenSSL, with opt-in via the inclusion of a separate package).

In almost all cases, it is an antipattern to use these features outside of high-level packages that are designed for curation. If a feature is optional, it can almost certainly be expressed as a separate package.

The [workspace] section

Packages can define a workspace which is a set of crates that will all share the same Cargo.lock and output directory. The [workspace] table can be defined as:

[workspace]

# Optional key, inferred from path dependencies if not present.
# Additional non-path dependencies that should be included must be given here.
# In particular, for a virtual manifest, all members have to be listed.
members = ["path/to/member1", "path/to/member2", "path/to/member3/*"]

# Optional key, empty if not present.
exclude = ["path1", "path/to/dir2"]

Workspaces were added to Cargo as part of RFC 1525 and have a number of properties:

• A workspace can contain multiple crates where one of them is the root crate.
• The root crate's Cargo.toml contains the [workspace] table, but is not required to have other configuration.
• Whenever any crate in the workspace is compiled, output is placed in the workspace root (i.e., next to the root crate's Cargo.toml).
• The lock file for all crates in the workspace resides in the workspace root.
• The [patch], [replace] and [profile.*] sections in Cargo.toml are only recognized in the root crate's manifest, and ignored in member crates' manifests.

The root crate of a workspace, indicated by the presence of [workspace] in its manifest, is responsible for defining the entire workspace. All path dependencies residing in the workspace directory become members. You can add additional packages to the workspace by listing them in the members key. Note that members of the workspaces listed explicitly will also have their path dependencies included in the workspace. Sometimes a package may have a lot of workspace members and it can be onerous to keep up to date. The members list can also use globs to match multiple paths. Finally, the exclude key can be used to blacklist paths from being included in a workspace. This can be useful if some path dependencies aren't desired to be in the workspace at all.

The package.workspace manifest key (described above) is used in member crates to point at a workspace's root crate. If this key is omitted then it is inferred to be the first crate whose manifest contains [workspace] upwards in the filesystem.

A crate may either specify package.workspace or specify [workspace]. That is, a crate cannot both be a root crate in a workspace (contain [workspace]) and also be a member crate of another workspace (contain package.workspace).

Most of the time workspaces will not need to be dealt with as cargo new and cargo init will handle workspace configuration automatically.

Virtual Manifest

In workspace manifests, if the package table is present, the workspace root crate will be treated as a normal package, as well as a workspace. If the package table is not present in a workspace manifest, it is called a virtual manifest.

Package selection

In a workspace, package-related cargo commands like cargo build apply to packages selected by -p / --package or --workspace command-line parameters. When neither is specified, the optional default-members configuration is used:

[workspace]
members = ["path/to/member1", "path/to/member2", "path/to/member3/*"]
default-members = ["path/to/member2", "path/to/member3/foo"]

When specified, default-members must expand to a subset of members.

When default-members is not specified, the default is the root manifest if it is a package, or every member manifest (as if --workspace were specified on the command-line) for virtual workspaces.

The project layout

If your package is an executable, name the main source file src/main.rs. If it is a library, name the main source file src/lib.rs.

Cargo will also treat any files located in src/bin/*.rs as executables. If your executable consists of more than just one source file, you might also use a directory inside src/bin containing a main.rs file which will be treated as an executable with a name of the parent directory.

Your package can optionally contain folders named examples, tests, and benches, which Cargo will treat as containing examples, integration tests, and benchmarks respectively. Analogous to bin targets, they may be composed of single files or directories with a main.rs file.

▾ src/           # directory containing source files
  lib.rs         # the main entry point for libraries and packages
  main.rs        # the main entry point for packages producing executables
  ▾ bin/         # (optional) directory containing additional executables
    *.rs
  ▾ */           # (optional) directories containing multi-file executables
    main.rs
▾ examples/      # (optional) examples
  *.rs
  ▾ */           # (optional) directories containing multi-file examples
    main.rs
▾ tests/         # (optional) integration tests
  *.rs
  ▾ */           # (optional) directories containing multi-file tests
    main.rs
▾ benches/       # (optional) benchmarks
  *.rs
  ▾ */           # (optional) directories containing multi-file benchmarks
    main.rs

To structure your code after you've created the files and folders for your package, you should remember to use Rust's module system, which you can read about in the book.

See Configuring a target below for more details on manually configuring target settings. See Target auto-discovery below for more information on controlling how Cargo automatically infers targets.

Examples

Files located under examples are example uses of the functionality provided by the library. When compiled, they are placed in the target/examples directory.

They can compile either as executables (with a main() function) or libraries and pull in the library by using extern crate <library-name>. They are compiled when you run your tests to protect them from bitrotting.

You can run individual executable examples with the command cargo run --example <example-name>.

Specify crate-type to make an example be compiled as a library (additional information about crate types is available in The Rust Reference):

[[example]]
name = "foo"
crate-type = ["staticlib"]

You can build individual library examples with the command cargo build --example <example-name>.

Tests

When you run cargo test, Cargo will:

• compile and run your library’s unit tests, which are in the files reachable from lib.rs (naturally, any sections marked with #[cfg(test)] will be considered at this stage);
• compile and run your library’s documentation tests, which are embedded inside of documentation blocks;
• compile and run your library’s integration tests; and
• compile your library’s examples.

Integration tests

Each file in tests/*.rs is an integration test. When you run cargo test, Cargo will compile each of these files as a separate crate. The crate can link to your library by using extern crate <library-name>, like any other code that depends on it.

Cargo will not automatically compile files inside subdirectories of tests, but an integration test can import modules from these directories as usual. For example, if you want several integration tests to share some code, you can put the shared code in tests/common/mod.rs and then put mod common; in each of the test files.

Configuring a target

All of the [[bin]], [lib], [[bench]], [[test]], and [[example]] sections support similar configuration for specifying how a target should be built. The double-bracket sections like [[bin]] are array-of-table of TOML, which means you can write more than one [[bin]] section to make several executables in your crate.

The example below uses [lib], but it also applies to all other sections as well. All values listed are the defaults for that option unless otherwise specified.

[package]
# ...

[lib]
# The name of a target is the name of the library that will be generated. This
# is defaulted to the name of the package, with any dashes replaced
# with underscores. (Rust `extern crate` declarations reference this name;
# therefore the value must be a valid Rust identifier to be usable.)
name = "foo"

# This field points at where the crate is located, relative to the `Cargo.toml`.
path = "src/lib.rs"

# A flag for enabling unit tests for this target. This is used by `cargo test`.
test = true

# A flag for enabling documentation tests for this target. This is only relevant
# for libraries, it has no effect on other sections. This is used by
# `cargo test`.
doctest = true

# A flag for enabling benchmarks for this target. This is used by `cargo bench`.
bench = true

# A flag for enabling documentation of this target. This is used by `cargo doc`.
doc = true

# If the target is meant to be a compiler plugin, this field must be set to true
# for Cargo to correctly compile it and make it available for all dependencies.
plugin = false

# If the target is meant to be a "macros 1.1" procedural macro, this field must
# be set to true.
proc-macro = false

# If set to false, `cargo test` will omit the `--test` flag to rustc, which
# stops it from generating a test harness. This is useful when the binary being
# built manages the test runner itself.
harness = true

# If set then a target can be configured to use a different edition than the
# `[package]` is configured to use, perhaps only compiling a library with the
# 2018 edition or only compiling one unit test with the 2015 edition. By default
# all targets are compiled with the edition specified in `[package]`.
edition = '2015'

# Here's an example of a TOML "array of tables" section, in this case specifying
# a binary target name and path.
[[bin]]
name = "my-cool-binary"
path = "src/my-cool-binary.rs"

Target auto-discovery

By default, Cargo automatically determines the targets to build based on the layout of the files on the filesystem. The target configuration tables, such as [lib], [[bin]], [[test]], [[bench]], or [[example]], can be used to add additional targets that don't follow the standard directory layout.

The automatic target discovery can be disabled so that only manually configured targets will be built. Setting the keys autobins, autoexamples, autotests, or autobenches to false in the [package] section will disable auto-discovery of the corresponding target type.

Disabling automatic discovery should only be needed for specialized situations. For example, if you have a library where you want a module named bin, this would present a problem because Cargo would usually attempt to compile anything in the bin directory as an executable. Here is a sample layout of this scenario:

├── Cargo.toml
└── src
    ├── lib.rs
    └── bin
        └── mod.rs

To prevent Cargo from inferring src/bin/mod.rs as an executable, set autobins = false in Cargo.toml to disable auto-discovery:

[package]
# …
autobins = false

Note: For packages with the 2015 edition, the default for auto-discovery is false if at least one target is manually defined in Cargo.toml. Beginning with the 2018 edition, the default is always true.

The required-features field (optional)

The required-features field specifies which features the target needs in order to be built. If any of the required features are not selected, the target will be skipped. This is only relevant for the [[bin]], [[bench]], [[test]], and [[example]] sections, it has no effect on [lib].

[features]
# ...
postgres = []
sqlite = []
tools = []

[[bin]]
# ...
required-features = ["postgres", "tools"]

Building dynamic or static libraries

If your package produces a library, you can specify which kind of library to build by explicitly listing the library in your Cargo.toml:

# ...

[lib]
name = "..."
crate-type = ["dylib"] # could be `staticlib` as well

The available options are dylib, rlib, staticlib, cdylib, and proc-macro.

You can read more about the different crate types in the Rust Reference Manual

The [patch] Section

This section of Cargo.toml can be used to override dependencies with other copies. The syntax is similar to the [dependencies] section:

[patch.crates-io]
foo = { git = 'https://github.com/example/foo' }
bar = { path = 'my/local/bar' }

[dependencies.baz]
git = 'https://github.com/example/baz'

[patch.'https://github.com/example/baz']
baz = { git = 'https://github.com/example/patched-baz', branch = 'my-branch' }

The [patch] table is made of dependency-like sub-tables. Each key after [patch] is a URL of the source that is being patched, or the name of a registry. The name crates-io may be used to override the default registry crates.io. The first [patch] in the example above demonstrates overriding crates.io, and the second [patch] demonstrates overriding a git source.

Each entry in these tables is a normal dependency specification, the same as found in the [dependencies] section of the manifest. The dependencies listed in the [patch] section are resolved and used to patch the source at the URL specified. The above manifest snippet patches the crates-io source (e.g. crates.io itself) with the foo crate and bar crate. It also patches the https://github.com/example/baz source with a my-branch that comes from elsewhere.

Sources can be patched with versions of crates that do not exist, and they can also be patched with versions of crates that already exist. If a source is patched with a crate version that already exists in the source, then the source's original crate is replaced.

More information about overriding dependencies can be found in the overriding dependencies section of the documentation and RFC 1969 for the technical specification of this feature.

Using [patch] with multiple versions

You can patch in multiple versions of the same crate with the package key used to rename dependencies. For example let's say that the serde crate has a bugfix that we'd like to use to its 1.* series but we'd also like to prototype using a 2.0.0 version of serde we have in our git repository. To configure this we'd do:

[patch.crates-io]
serde = { git = 'https://github.com/serde-rs/serde' }
serde2 = { git = 'https://github.com/example/serde', package = 'serde', branch = 'v2' }

The first serde = ... directive indicates that serde 1.* should be used from the git repository (pulling in the bugfix we need) and the second serde2 = ... directive indicates that the serde package should also be pulled from the v2 branch of https://github.com/example/serde. We're assuming here that Cargo.toml on that branch mentions version 2.0.0.

Note that when using the package key the serde2 identifier here is actually ignored. We simply need a unique name which doesn't conflict with other patched crates.

The [replace] Section

This section of Cargo.toml can be used to override dependencies with other copies. The syntax is similar to the [dependencies] section:

[replace]
"foo:0.1.0" = { git = 'https://github.com/example/foo' }
"bar:1.0.2" = { path = 'my/local/bar' }

Each key in the [replace] table is a package ID specification, which allows arbitrarily choosing a node in the dependency graph to override. The value of each key is the same as the [dependencies] syntax for specifying dependencies, except that you can't specify features. Note that when a crate is overridden the copy it's overridden with must have both the same name and version, but it can come from a different source (e.g., git or a local path).

More information about overriding dependencies can be found in the overriding dependencies section of the documentation.

Configuration

This document will explain how Cargo’s configuration system works, as well as available keys or configuration. For configuration of a package through its manifest, see the manifest format.

Hierarchical structure

Cargo allows local configuration for a particular package as well as global configuration, like git. Cargo extends this to a hierarchical strategy. If, for example, Cargo were invoked in /projects/foo/bar/baz, then the following configuration files would be probed for and unified in this order:

• /projects/foo/bar/baz/.cargo/config
• /projects/foo/bar/.cargo/config
• /projects/foo/.cargo/config
• /projects/.cargo/config
• /.cargo/config
• $CARGO_HOME/config ($CARGO_HOME defaults to $HOME/.cargo)

With this structure, you can specify configuration per-package, and even possibly check it into version control. You can also specify personal defaults with a configuration file in your home directory.

Configuration format

All configuration is currently in the TOML format (like the manifest), with simple key-value pairs inside of sections (tables) which all get merged together.

Configuration keys

All of the following keys are optional, and their defaults are listed as their value unless otherwise noted.

Key values that specify a tool may be given as an absolute path, a relative path or as a pathless tool name. Absolute paths and pathless tool names are used as given. Relative paths are resolved relative to the parent directory of the .cargo directory of the config file that the value resides within.

# An array of paths to local repositories which are to be used as overrides for
# dependencies. For more information see the Specifying Dependencies guide.
paths = ["/path/to/override"]

[cargo-new]
# This is your name/email to place in the `authors` section of a new Cargo.toml
# that is generated. If not present, then `git` will be probed, and if that is
# not present then `$USER` and `$EMAIL` will be used.
name = "..."
email = "..."

# By default `cargo new` will initialize a new Git repository. This key can
# be set to change the version control system used. Valid values are `git`,
# `hg` (for Mercurial), `pijul`, `fossil`, or `none` to disable this behavior.
vcs = "none"

# For the following sections, $triple refers to any valid target triple, not the
# literal string "$triple", and it will apply whenever that target triple is
# being compiled to. 'cfg(...)' refers to the Rust-like `#[cfg]` syntax for
# conditional compilation.
[target.$triple]
# This is the linker which is passed to rustc (via `-C linker=`) when the `$triple`
# is being compiled for. By default this flag is not passed to the compiler.
linker = ".."
# Same but for the library archiver which is passed to rustc via `-C ar=`.
ar = ".."
# If a runner is provided, compiled targets for the `$triple` will be executed
# by invoking the specified runner executable with actual target as first argument.
# This applies to `cargo run`, `cargo test` and `cargo bench` commands.
# By default compiled targets are executed directly.
runner = ".."
# custom flags to pass to all compiler invocations that target $triple
# this value overrides build.rustflags when both are present
rustflags = ["..", ".."]

[target.'cfg(...)']
# Similar for the $triple configuration, but using the `cfg` syntax.
# If several `cfg` and $triple targets are candidates, then the rustflags
# are concatenated. The `cfg` syntax only applies to rustflags, and not to
# linker.
rustflags = ["..", ".."]
# Similar for the $triple configuration, but using the `cfg` syntax.
# If one or more `cfg`s, and a $triple target are candidates, then the $triple
# will be used
# If several `cfg` are candidates, then the build will error
runner = ".."

# Configuration keys related to the registry
[registry]
index = "..."   # URL of the registry index (defaults to the index of crates.io)
default = "..." # Name of the default registry to use (can be overridden with
                # --registry)

# Configuration keys for registries other than crates.io.
# `$name` should be the name of the registry, which will be used for
# dependencies in `Cargo.toml` files and the `--registry` command-line flag.
# Registry names should only contain alphanumeric characters, `-`, or `_`.
[registries.$name]
index = "..."   # URL of the registry index

[http]
proxy = "host:port" # HTTP proxy to use for HTTP requests (defaults to none)
                    # in libcurl format, e.g., "socks5h://host:port"
timeout = 30        # Timeout for each HTTP request, in seconds
cainfo = "cert.pem" # Path to Certificate Authority (CA) bundle (optional)
check-revoke = true # Indicates whether SSL certs are checked for revocation
low-speed-limit = 5 # Lower threshold for bytes/sec (10 = default, 0 = disabled)
multiplexing = true # whether or not to use HTTP/2 multiplexing where possible

# This setting can be used to help debug what's going on with HTTP requests made
# by Cargo. When set to `true` then Cargo's normal debug logging will be filled
# in with HTTP information, which you can extract with
# `CARGO_LOG=cargo::ops::registry=debug` (and `trace` may print more).
#
# Be wary when posting these logs elsewhere though, it may be the case that a
# header has an authentication token in it you don't want leaked! Be sure to
# briefly review logs before posting them.
debug = false

[build]
jobs = 1                  # number of parallel jobs, defaults to # of CPUs
rustc = "rustc"           # the rust compiler tool
rustdoc = "rustdoc"       # the doc generator tool
target = "triple"         # build for the target triple (ignored by `cargo install`)
target-dir = "target"     # path of where to place all generated artifacts
rustflags = ["..", ".."]  # custom flags to pass to all compiler invocations
rustdocflags = ["..", ".."]  # custom flags to pass to rustdoc
incremental = true        # whether or not to enable incremental compilation
                          # If `incremental` is not set, then the value from
                          # the profile is used.
dep-info-basedir = ".."   # full path for the base directory for targets in depfiles

[term]
verbose = false        # whether cargo provides verbose output
color = 'auto'         # whether cargo colorizes output

# Network configuration
[net]
retry = 2 # number of times a network call will automatically retried
git-fetch-with-cli = false  # if `true` we'll use `git`-the-CLI to fetch git repos
offline = false # do not access the network, but otherwise try to proceed if possible

# Alias cargo commands. The first 4 aliases are built in. If your
# command requires grouped whitespace use the list format.
[alias]
b = "build"
c = "check"
t = "test"
r = "run"
rr = "run --release"
space_example = ["run", "--release", "--", "\"command list\""]

Environment variables

Cargo can also be configured through environment variables in addition to the TOML syntax above. For each configuration key above of the form foo.bar the environment variable CARGO_FOO_BAR can also be used to define the value. For example the build.jobs key can also be defined by CARGO_BUILD_JOBS.

Environment variables will take precedent over TOML configuration, and currently only integer, boolean, and string keys are supported to be defined by environment variables. This means that source replacement, which is expressed by tables, cannot be configured through environment variables.

In addition to the system above, Cargo recognizes a few other specific environment variables.

Credentials

Configuration values with sensitive information are stored in the $CARGO_HOME/credentials file. This file is automatically created and updated by cargo login. It follows the same format as Cargo config files.

[registry]
token = "..."   # Access token for crates.io

# `$name` should be a registry name (see above for more information about
# configuring registries).
[registries.$name]
token = "..."   # Access token for the named registry

Tokens are used by some Cargo commands such as cargo publish for authenticating with remote registries. Care should be taken to protect the tokens and to keep them secret.

As with most other config values, tokens may be specified with environment variables. The token for crates.io may be specified with the CARGO_REGISTRY_TOKEN environment variable. Tokens for other registries may be specified with environment variables of the form CARGO_REGISTRIES_NAME_TOKEN where NAME is the name of the registry in all capital letters.

Environment Variables

Cargo sets and reads a number of environment variables which your code can detect or override. Here is a list of the variables Cargo sets, organized by when it interacts with them:

Environment variables Cargo reads

You can override these environment variables to change Cargo's behavior on your system:

• CARGO_HOME — Cargo maintains a local cache of the registry index and of git checkouts of crates. By default these are stored under $HOME/.cargo, but this variable overrides the location of this directory. Once a crate is cached it is not removed by the clean command. For more details refer to the guide.
• CARGO_TARGET_DIR — Location of where to place all generated artifacts, relative to the current working directory.
• RUSTC — Instead of running rustc, Cargo will execute this specified compiler instead.
• RUSTC_WRAPPER — Instead of simply running rustc, Cargo will execute this specified wrapper instead, passing as its commandline arguments the rustc invocation, with the first argument being rustc.
• RUSTDOC — Instead of running rustdoc, Cargo will execute this specified rustdoc instance instead.
• RUSTDOCFLAGS — A space-separated list of custom flags to pass to all rustdoc invocations that Cargo performs. In contrast with cargo rustdoc, this is useful for passing a flag to all rustdoc instances.
• RUSTFLAGS — A space-separated list of custom flags to pass to all compiler invocations that Cargo performs. In contrast with cargo rustc, this is useful for passing a flag to all compiler instances.
• CARGO_INCREMENTAL — If this is set to 1 then Cargo will force incremental compilation to be enabled for the current compilation, and when set to 0 it will force disabling it. If this env var isn't present then cargo's defaults will otherwise be used.
• CARGO_CACHE_RUSTC_INFO — If this is set to 0 then Cargo will not try to cache compiler version information.

Note that Cargo will also read environment variables for .cargo/config configuration values, as described in that documentation

Environment variables Cargo sets for crates

Cargo exposes these environment variables to your crate when it is compiled. Note that this applies for test binaries as well. To get the value of any of these variables in a Rust program, do this:

# #![allow(unused_variables)]
#fn main() {
let version = env!("CARGO_PKG_VERSION");
#}

version will now contain the value of CARGO_PKG_VERSION.

• CARGO - Path to the cargo binary performing the build.
• CARGO_MANIFEST_DIR - The directory containing the manifest of your package.
• CARGO_PKG_VERSION - The full version of your package.
• CARGO_PKG_VERSION_MAJOR - The major version of your package.
• CARGO_PKG_VERSION_MINOR - The minor version of your package.
• CARGO_PKG_VERSION_PATCH - The patch version of your package.
• CARGO_PKG_VERSION_PRE - The pre-release version of your package.
• CARGO_PKG_AUTHORS - Colon separated list of authors from the manifest of your package.
• CARGO_PKG_NAME - The name of your package.
• CARGO_PKG_DESCRIPTION - The description from the manifest of your package.
• CARGO_PKG_HOMEPAGE - The home page from the manifest of your package.
• CARGO_PKG_REPOSITORY - The repository from the manifest of your package.
• OUT_DIR - If the package has a build script, this is set to the folder where the build script should place its output. See below for more information.

Environment variables Cargo sets for build scripts

Cargo sets several environment variables when build scripts are run. Because these variables are not yet set when the build script is compiled, the above example using env! won't work and instead you'll need to retrieve the values when the build script is run:

# #![allow(unused_variables)]
#fn main() {
use std::env;
let out_dir = env::var("OUT_DIR").unwrap();
#}

out_dir will now contain the value of OUT_DIR.

• CARGO - Path to the cargo binary performing the build.
• CARGO_MANIFEST_DIR - The directory containing the manifest for the package being built (the package containing the build script). Also note that this is the value of the current working directory of the build script when it starts.
• CARGO_MANIFEST_LINKS - the manifest links value.
• CARGO_FEATURE_<name> - For each activated feature of the package being built, this environment variable will be present where <name> is the name of the feature uppercased and having - translated to _.
• CARGO_CFG_<cfg> - For each configuration option of the package being built, this environment variable will contain the value of the configuration, where <cfg> is the name of the configuration uppercased and having - translated to _. Boolean configurations are present if they are set, and not present otherwise. Configurations with multiple values are joined to a single variable with the values delimited by ,.
• OUT_DIR - the folder in which all output should be placed. This folder is inside the build directory for the package being built, and it is unique for the package in question.
• TARGET - the target triple that is being compiled for. Native code should be compiled for this triple. See the Target Triple description for more information.
• HOST - the host triple of the rust compiler.
• NUM_JOBS - the parallelism specified as the top-level parallelism. This can be useful to pass a -j parameter to a system like make. Note that care should be taken when interpreting this environment variable. For historical purposes this is still provided but recent versions of Cargo, for example, do not need to run make -j as it'll automatically happen. Cargo implements its own jobserver and will allow build scripts to inherit this information, so programs compatible with GNU make jobservers will already have appropriately configured parallelism.
• OPT_LEVEL, DEBUG - values of the corresponding variables for the profile currently being built.
• PROFILE - release for release builds, debug for other builds.
• DEP_<name>_<key> - For more information about this set of environment variables, see build script documentation about links.
• RUSTC, RUSTDOC - the compiler and documentation generator that Cargo has resolved to use, passed to the build script so it might use it as well.
• RUSTC_LINKER - The path to the linker binary that Cargo has resolved to use for the current target, if specified. The linker can be changed by editing .cargo/config; see the documentation about cargo configuration for more information.

Environment variables Cargo sets for 3rd party subcommands

Cargo exposes this environment variable to 3rd party subcommands (ie. programs named cargo-foobar placed in $PATH):

• CARGO - Path to the cargo binary performing the build.

Build Scripts

Some packages need to compile third-party non-Rust code, for example C libraries. Other packages need to link to C libraries which can either be located on the system or possibly need to be built from source. Others still need facilities for functionality such as code generation before building (think parser generators).

Cargo does not aim to replace other tools that are well-optimized for these tasks, but it does integrate with them with the build configuration option.

[package]
# ...
build = "build.rs"

The Rust file designated by the build command (relative to the package root) will be compiled and invoked before anything else is compiled in the package, allowing your Rust code to depend on the built or generated artifacts. By default Cargo looks for a "build.rs" file in a package root (even if you do not specify a value for build). Use build = "custom_build_name.rs" to specify a custom build name or build = false to disable automatic detection of the build script.

Some example use cases of the build command are:

• Building a bundled C library.
• Finding a C library on the host system.
• Generating a Rust module from a specification.
• Performing any platform-specific configuration needed for the crate.

Each of these use cases will be detailed in full below to give examples of how the build command works.

Inputs to the Build Script

When the build script is run, there are a number of inputs to the build script, all passed in the form of environment variables.

In addition to environment variables, the build script’s current directory is the source directory of the build script’s package.

Outputs of the Build Script

All the lines printed to stdout by a build script are written to a file like target/debug/build/<pkg>/output (the precise location may depend on your configuration). If you would like to see such output directly in your terminal, invoke cargo as 'very verbose' with the -vv flag. Note that if neither the build script nor package source files are modified, subsequent calls to cargo with -vv will not print output to the terminal because a new build is not executed. Run cargo clean before each cargo invocation if you want to ensure that output is always displayed on your terminal. Any line that starts with cargo: is interpreted directly by Cargo. This line must be of the form cargo:key=value, like the examples below:

# specially recognized by Cargo
cargo:rustc-link-lib=static=foo
cargo:rustc-link-search=native=/path/to/foo
cargo:rustc-cfg=foo
cargo:rustc-env=FOO=bar
cargo:rustc-cdylib-link-arg=-Wl,-soname,libfoo.so.1.2.3
# arbitrary user-defined metadata
cargo:root=/path/to/foo
cargo:libdir=/path/to/foo/lib
cargo:include=/path/to/foo/include

On the other hand, lines printed to stderr are written to a file like target/debug/build/<pkg>/stderr but are not interpreted by cargo.

There are a few special keys that Cargo recognizes, some affecting how the crate is built:

• rustc-link-lib=[KIND=]NAME indicates that the specified value is a library name and should be passed to the compiler as a -l flag. The optional KIND can be one of static, dylib (the default), or framework, see rustc --help for more details.

• rustc-link-search=[KIND=]PATH indicates the specified value is a library search path and should be passed to the compiler as a -L flag. The optional KIND can be one of dependency, crate, native, framework or all (the default), see rustc --help for more details.

• rustc-flags=FLAGS is a set of flags passed to the compiler, only -l and -L flags are supported.

• rustc-cfg=FEATURE indicates that the specified feature will be passed as a --cfg flag to the compiler. This is often useful for performing compile-time detection of various features.

• rustc-env=VAR=VALUE indicates that the specified environment variable will be added to the environment which the compiler is run within. The value can be then retrieved by the env! macro in the compiled crate. This is useful for embedding additional metadata in crate's code, such as the hash of Git HEAD or the unique identifier of a continuous integration server.

• rustc-cdylib-link-arg=FLAG is a flag passed to the compiler as -C link-arg=FLAG when building a cdylib. Its usage is highly platform specific. It is useful to set the shared library version or the runtime-path.

• rerun-if-changed=PATH is a path to a file or directory which indicates that the build script should be re-run if it changes (detected by a more-recent last-modified timestamp on the file). Normally build scripts are re-run if any file inside the crate root changes, but this can be used to scope changes to just a small set of files. (If this path points to a directory the entire directory will not be traversed for changes -- only changes to the timestamp of the directory itself (which corresponds to some types of changes within the directory, depending on platform) will trigger a rebuild. To request a re-run on any changes within an entire directory, print a line for the directory and another line for everything inside it, recursively.) Note that if the build script itself (or one of its dependencies) changes, then it's rebuilt and rerun unconditionally, so cargo:rerun-if-changed=build.rs is almost always redundant (unless you want to ignore changes in all other files except for build.rs).

• rerun-if-env-changed=VAR is the name of an environment variable which indicates that if the environment variable's value changes the build script should be rerun. This basically behaves the same as rerun-if-changed except that it works with environment variables instead. Note that the environment variables here are intended for global environment variables like CC and such, it's not necessary to use this for env vars like TARGET that Cargo sets. Also note that if rerun-if-env-changed is printed out then Cargo will only rerun the build script if those environment variables change or if files printed out by rerun-if-changed change.

• warning=MESSAGE is a message that will be printed to the main console after a build script has finished running. Warnings are only shown for path dependencies (that is, those you're working on locally), so for example warnings printed out in crates.io crates are not emitted by default.

Any other element is a user-defined metadata that will be passed to dependents. More information about this can be found in the links section.

Build Dependencies

Build scripts are also allowed to have dependencies on other Cargo-based crates. Dependencies are declared through the build-dependencies section of the manifest.

[build-dependencies]
foo = { git = "https://github.com/your-packages/foo" }

The build script does not have access to the dependencies listed in the dependencies or dev-dependencies section (they’re not built yet!). All build dependencies will also not be available to the package itself unless explicitly stated as so.

The links Manifest Key

In addition to the manifest key build, Cargo also supports a links manifest key to declare the name of a native library that is being linked to:

[package]
# ...
links = "foo"
build = "build.rs"

This manifest states that the package links to the libfoo native library, and it also has a build script for locating and/or building the library. Cargo requires that a build command is specified if a links entry is also specified.

The purpose of this manifest key is to give Cargo an understanding about the set of native dependencies that a package has, as well as providing a principled system of passing metadata between package build scripts.

Primarily, Cargo requires that there is at most one package per links value. In other words, it’s forbidden to have two packages link to the same native library. Note, however, that there are conventions in place to alleviate this.

As mentioned above in the output format, each build script can generate an arbitrary set of metadata in the form of key-value pairs. This metadata is passed to the build scripts of dependent packages. For example, if libbar depends on libfoo, then if libfoo generates key=value as part of its metadata, then the build script of libbar will have the environment variables DEP_FOO_KEY=value.

Note that metadata is only passed to immediate dependents, not transitive dependents. The motivation for this metadata passing is outlined in the linking to system libraries case study below.

Overriding Build Scripts

If a manifest contains a links key, then Cargo supports overriding the build script specified with a custom library. The purpose of this functionality is to prevent running the build script in question altogether and instead supply the metadata ahead of time.

To override a build script, place the following configuration in any acceptable Cargo configuration location.

[target.x86_64-unknown-linux-gnu.foo]
rustc-link-search = ["/path/to/foo"]
rustc-link-lib = ["foo"]
root = "/path/to/foo"
key = "value"

This section states that for the target x86_64-unknown-linux-gnu the library named foo has the metadata specified. This metadata is the same as the metadata generated as if the build script had run, providing a number of key/value pairs where the rustc-flags, rustc-link-search, and rustc-link-lib keys are slightly special.

With this configuration, if a package declares that it links to foo then the build script will not be compiled or run, and the metadata specified will instead be used.

Case study: Code generation

Some Cargo packages need to have code generated just before they are compiled for various reasons. Here we’ll walk through a simple example which generates a library call as part of the build script.

First, let’s take a look at the directory structure of this package:

.
├── Cargo.toml
├── build.rs
└── src
    └── main.rs

1 directory, 3 files

Here we can see that we have a build.rs build script and our binary in main.rs. Next, let’s take a look at the manifest:

# Cargo.toml

[package]
name = "hello-from-generated-code"
version = "0.1.0"
authors = ["you@example.com"]
build = "build.rs"

Here we can see we’ve got a build script specified which we’ll use to generate some code. Let’s see what’s inside the build script:

// build.rs

use std::env;
use std::fs::File;
use std::io::Write;
use std::path::Path;

fn main() {
    let out_dir = env::var("OUT_DIR").unwrap();
    let dest_path = Path::new(&out_dir).join("hello.rs");
    let mut f = File::create(&dest_path).unwrap();

    f.write_all(b"
        pub fn message() -> &'static str {
            \"Hello, World!\"
        }
    ").unwrap();
}

There’s a couple of points of note here:

• The script uses the OUT_DIR environment variable to discover where the output files should be located. It can use the process’ current working directory to find where the input files should be located, but in this case we don’t have any input files.
• In general, build scripts should not modify any files outside of OUT_DIR. It may seem fine on the first blush, but it does cause problems when you use such crate as a dependency, because there's an implicit invariant that sources in .cargo/registry should be immutable. cargo won't allow such scripts when packaging.
• This script is relatively simple as it just writes out a small generated file. One could imagine that other more fanciful operations could take place such as generating a Rust module from a C header file or another language definition, for example.

Next, let’s peek at the library itself:

// src/main.rs

include!(concat!(env!("OUT_DIR"), "/hello.rs"));

fn main() {
    println!("{}", message());
}

This is where the real magic happens. The library is using the rustc-defined include! macro in combination with the concat! and env! macros to include the generated file (hello.rs) into the crate’s compilation.

Using the structure shown here, crates can include any number of generated files from the build script itself.

Case study: Building some native code

Sometimes it’s necessary to build some native C or C++ code as part of a package. This is another excellent use case of leveraging the build script to build a native library before the Rust crate itself. As an example, we’ll create a Rust library which calls into C to print “Hello, World!”.

Like above, let’s first take a look at the package layout:

.
├── Cargo.toml
├── build.rs
└── src
    ├── hello.c
    └── main.rs

1 directory, 4 files

Pretty similar to before! Next, the manifest:

# Cargo.toml

[package]
name = "hello-world-from-c"
version = "0.1.0"
authors = ["you@example.com"]
build = "build.rs"

For now we’re not going to use any build dependencies, so let’s take a look at the build script now:

// build.rs

use std::process::Command;
use std::env;
use std::path::Path;

fn main() {
    let out_dir = env::var("OUT_DIR").unwrap();

    // note that there are a number of downsides to this approach, the comments
    // below detail how to improve the portability of these commands.
    Command::new("gcc").args(&["src/hello.c", "-c", "-fPIC", "-o"])
                       .arg(&format!("{}/hello.o", out_dir))
                       .status().unwrap();
    Command::new("ar").args(&["crus", "libhello.a", "hello.o"])
                      .current_dir(&Path::new(&out_dir))
                      .status().unwrap();

    println!("cargo:rustc-link-search=native={}", out_dir);
    println!("cargo:rustc-link-lib=static=hello");
}

This build script starts out by compiling our C file into an object file (by invoking gcc) and then converting this object file into a static library (by invoking ar). The final step is feedback to Cargo itself to say that our output was in out_dir and the compiler should link the crate to libhello.a statically via the -l static=hello flag.

Note that there are a number of drawbacks to this hardcoded approach:

• The gcc command itself is not portable across platforms. For example it’s unlikely that Windows platforms have gcc, and not even all Unix platforms may have gcc. The ar command is also in a similar situation.
• These commands do not take cross-compilation into account. If we’re cross compiling for a platform such as Android it’s unlikely that gcc will produce an ARM executable.

Not to fear, though, this is where a build-dependencies entry would help! The Cargo ecosystem has a number of packages to make this sort of task much easier, portable, and standardized. For example, the build script could be written as:

// build.rs

// Bring in a dependency on an externally maintained `cc` package which manages
// invoking the C compiler.
extern crate cc;

fn main() {
    cc::Build::new()
        .file("src/hello.c")
        .compile("hello");
}

Add a build time dependency on the cc crate with the following addition to your Cargo.toml:

[build-dependencies]
cc = "1.0"

The cc crate abstracts a range of build script requirements for C code:

• It invokes the appropriate compiler (MSVC for windows, gcc for MinGW, cc for Unix platforms, etc.).
• It takes the TARGET variable into account by passing appropriate flags to the compiler being used.
• Other environment variables, such as OPT_LEVEL, DEBUG, etc., are all handled automatically.
• The stdout output and OUT_DIR locations are also handled by the cc library.

Here we can start to see some of the major benefits of farming as much functionality as possible out to common build dependencies rather than duplicating logic across all build scripts!

Back to the case study though, let’s take a quick look at the contents of the src directory:

// src/hello.c

#include <stdio.h>

void hello() {
    printf("Hello, World!\n");
}

// src/main.rs

// Note the lack of the `#[link]` attribute. We’re delegating the responsibility
// of selecting what to link to over to the build script rather than hardcoding
// it in the source file.
extern { fn hello(); }

fn main() {
    unsafe { hello(); }
}

And there we go! This should complete our example of building some C code from a Cargo package using the build script itself. This also shows why using a build dependency can be crucial in many situations and even much more concise!

We’ve also seen a brief example of how a build script can use a crate as a dependency purely for the build process and not for the crate itself at runtime.

Case study: Linking to system libraries

The final case study here will be investigating how a Cargo library links to a system library and how the build script is leveraged to support this use case.

Quite frequently a Rust crate wants to link to a native library often provided on the system to bind its functionality or just use it as part of an implementation detail. This is quite a nuanced problem when it comes to performing this in a platform-agnostic fashion, and the purpose of a build script is again to farm out as much of this as possible to make this as easy as possible for consumers.

As an example to follow, let’s take a look at one of Cargo’s own dependencies, libgit2. The C library has a number of constraints:

• It has an optional dependency on OpenSSL on Unix to implement the https transport.
• It has an optional dependency on libssh2 on all platforms to implement the ssh transport.
• It is often not installed on all systems by default.
• It can be built from source using cmake.

To visualize what’s going on here, let’s take a look at the manifest for the relevant Cargo package that links to the native C library.

[package]
name = "libgit2-sys"
version = "0.1.0"
authors = ["..."]
links = "git2"
build = "build.rs"

[dependencies]
libssh2-sys = { git = "https://github.com/alexcrichton/ssh2-rs" }

[target.'cfg(unix)'.dependencies]
openssl-sys = { git = "https://github.com/alexcrichton/openssl-sys" }

# ...

As the above manifests show, we’ve got a build script specified, but it’s worth noting that this example has a links entry which indicates that the crate (libgit2-sys) links to the git2 native library.

Here we also see that we chose to have the Rust crate have an unconditional dependency on libssh2 via the libssh2-sys crate, as well as a platform-specific dependency on openssl-sys for *nix (other variants elided for now). It may seem a little counterintuitive to express C dependencies in the Cargo manifest, but this is actually using one of Cargo’s conventions in this space.

*-sys Packages

To alleviate linking to system libraries, crates.io has a convention of package naming and functionality. Any package named foo-sys should provide two major pieces of functionality:

• The library crate should link to the native library libfoo. This will often probe the current system for libfoo before resorting to building from source.
• The library crate should provide declarations for functions in libfoo, but not bindings or higher-level abstractions.

The set of *-sys packages provides a common set of dependencies for linking to native libraries. There are a number of benefits earned from having this convention of native-library-related packages:

• Common dependencies on foo-sys alleviates the above rule about one package per value of links.
• A common dependency allows centralizing logic on discovering libfoo itself (or building it from source).
• These dependencies are easily overridable.

Building libgit2

Now that we’ve got libgit2’s dependencies sorted out, we need to actually write the build script. We’re not going to look at specific snippets of code here and instead only take a look at the high-level details of the build script of libgit2-sys. This is not recommending all packages follow this strategy, but rather just outlining one specific strategy.

The first step of the build script should do is to query whether libgit2 is already installed on the host system. To do this we’ll leverage the preexisting tool pkg-config (when its available). We’ll also use a build-dependencies section to refactor out all the pkg-config related code (or someone’s already done that!).

If pkg-config failed to find libgit2, or if pkg-config just wasn’t installed, the next step is to build libgit2 from bundled source code (distributed as part of libgit2-sys itself). There are a few nuances when doing so that we need to take into account, however:

• The build system of libgit2, cmake, needs to be able to find libgit2’s optional dependency of libssh2. We’re sure we’ve already built it (it’s a Cargo dependency), we just need to communicate this information. To do this we leverage the metadata format to communicate information between build scripts. In this example the libssh2 package printed out cargo:root=... to tell us where libssh2 is installed at, and we can then pass this along to cmake with the CMAKE_PREFIX_PATH environment variable.

• We’ll need to handle some CFLAGS values when compiling C code (and tell cmake about this). Some flags we may want to pass are -m64 for 64-bit code, -m32 for 32-bit code, or -fPIC for 64-bit code as well.

• Finally, we’ll invoke cmake to place all output into the OUT_DIR environment variable, and then we’ll print the necessary metadata to instruct rustc how to link to libgit2.

Most of the functionality of this build script is easily refactorable into common dependencies, so our build script isn’t quite as intimidating as this descriptions! In reality it’s expected that build scripts are quite succinct by farming logic such as above to build dependencies.

Publishing on crates.io

Once you've got a library that you'd like to share with the world, it's time to publish it on crates.io! Publishing a crate is when a specific version is uploaded to be hosted on crates.io.

Take care when publishing a crate, because a publish is permanent. The version can never be overwritten, and the code cannot be deleted. There is no limit to the number of versions which can be published, however.

Before your first publish

First thing’s first, you’ll need an account on crates.io to acquire an API token. To do so, visit the home page and log in via a GitHub account (required for now). After this, visit your Account Settings page and run the cargo login command specified.

$ cargo login abcdefghijklmnopqrstuvwxyz012345

This command will inform Cargo of your API token and store it locally in your ~/.cargo/credentials. Note that this token is a secret and should not be shared with anyone else. If it leaks for any reason, you should regenerate it immediately.

Before publishing a new crate

Keep in mind that crate names on crates.io are allocated on a first-come-first- serve basis. Once a crate name is taken, it cannot be used for another crate.

Check out the metadata you can specify in Cargo.toml to ensure your crate can be discovered more easily! Before publishing, make sure you have filled out the following fields:

• authors
• license or license-file
• description
• homepage
• documentation
• repository

It would also be a good idea to include some keywords and categories, though they are not required.

If you are publishing a library, you may also want to consult the Rust API Guidelines.

Packaging a crate

The next step is to package up your crate and upload it to crates.io. For this we’ll use the cargo publish subcommand. This command performs the following steps:

1. Perform some verification checks on your package.
2. Compress your source code into a .crate file.
3. Extract the .crate file into a temporary directory and verify that it compiles.
4. Upload the .crate file to crates.io.
5. The registry will perform some additional checks on the uploaded package before adding it.

It is recommended that you first run cargo publish --dry-run (or cargo package which is equivalent) to ensure there aren't any warnings or errors before publishing. This will perform the first three steps listed above.

$ cargo publish --dry-run

You can inspect the generated .crate file in the target/package directory. crates.io currently has a 10MB size limit on the .crate file. You may want to check the size of the .crate file to ensure you didn't accidentally package up large assets that are not required to build your package, such as test data, website documentation, or code generation. You can check which files are included with the following command:

$ cargo package --list

Cargo will automatically ignore files ignored by your version control system when packaging, but if you want to specify an extra set of files to ignore you can use the exclude key in the manifest:

[package]
# ...
exclude = [
    "public/assets/*",
    "videos/*",
]

If you’d rather explicitly list the files to include, Cargo also supports an include key, which if set, overrides the exclude key:

[package]
# ...
include = [
    "**/*.rs",
    "Cargo.toml",
]

Uploading the crate

When you are ready to publish, use the cargo publish command to upload to crates.io:

$ cargo publish

And that’s it, you’ve now published your first crate!

Publishing a new version of an existing crate

In order to release a new version, change the version value specified in your Cargo.toml manifest. Keep in mind the semver rules, and consult RFC 1105 for what constitutes a semver-breaking change. Then run cargo publish as described above to upload the new version.

Managing a crates.io-based crate

Management of crates is primarily done through the command line cargo tool rather than the crates.io web interface. For this, there are a few subcommands to manage a crate.

cargo yank

Occasions may arise where you publish a version of a crate that actually ends up being broken for one reason or another (syntax error, forgot to include a file, etc.). For situations such as this, Cargo supports a “yank” of a version of a crate.

$ cargo yank --vers 1.0.1
$ cargo yank --vers 1.0.1 --undo

A yank does not delete any code. This feature is not intended for deleting accidentally uploaded secrets, for example. If that happens, you must reset those secrets immediately.

The semantics of a yanked version are that no new dependencies can be created against that version, but all existing dependencies continue to work. One of the major goals of crates.io is to act as a permanent archive of crates that does not change over time, and allowing deletion of a version would go against this goal. Essentially a yank means that all packages with a Cargo.lock will not break, while any future Cargo.lock files generated will not list the yanked version.

cargo owner

A crate is often developed by more than one person, or the primary maintainer may change over time! The owner of a crate is the only person allowed to publish new versions of the crate, but an owner may designate additional owners.

$ cargo owner --add my-buddy
$ cargo owner --remove my-buddy
$ cargo owner --add github:rust-lang:owners
$ cargo owner --remove github:rust-lang:owners

The owner IDs given to these commands must be GitHub user names or GitHub teams.

If a user name is given to --add, that user is invited as a “named” owner, with full rights to the crate. In addition to being able to publish or yank versions of the crate, they have the ability to add or remove owners, including the owner that made them an owner. Needless to say, you shouldn’t make people you don’t fully trust into a named owner. In order to become a named owner, a user must have logged into crates.io previously.

If a team name is given to --add, that team is invited as a “team” owner, with restricted right to the crate. While they have permission to publish or yank versions of the crate, they do not have the ability to add or remove owners. In addition to being more convenient for managing groups of owners, teams are just a bit more secure against owners becoming malicious.

The syntax for teams is currently github:org:team (see examples above). In order to invite a team as an owner one must be a member of that team. No such restriction applies to removing a team as an owner.

GitHub permissions

Team membership is not something GitHub provides simple public access to, and it is likely for you to encounter the following message when working with them:

It looks like you don’t have permission to query a necessary property from GitHub to complete this request. You may need to re-authenticate on crates.io to grant permission to read GitHub org memberships. Just go to https://crates.io/login.

This is basically a catch-all for “you tried to query a team, and one of the five levels of membership access control denied this”. That is not an exaggeration. GitHub’s support for team access control is Enterprise Grade.

The most likely cause of this is simply that you last logged in before this feature was added. We originally requested no permissions from GitHub when authenticating users, because we didn’t actually ever use the user’s token for anything other than logging them in. However to query team membership on your behalf, we now require the read:org scope.

You are free to deny us this scope, and everything that worked before teams were introduced will keep working. However you will never be able to add a team as an owner, or publish a crate as a team owner. If you ever attempt to do this, you will get the error above. You may also see this error if you ever try to publish a crate that you don’t own at all, but otherwise happens to have a team.

If you ever change your mind, or just aren’t sure if crates.io has sufficient permission, you can always go to https://crates.io/login, which will prompt you for permission if crates.io doesn’t have all the scopes it would like to.

An additional barrier to querying GitHub is that the organization may be actively denying third party access. To check this, you can go to:

https://github.com/organizations/:org/settings/oauth_application_policy

where :org is the name of the organization (e.g., rust-lang). You may see something like:

Where you may choose to explicitly remove crates.io from your organization’s blacklist, or simply press the “Remove Restrictions” button to allow all third party applications to access this data.

Alternatively, when crates.io requested the read:org scope, you could have explicitly whitelisted crates.io querying the org in question by pressing the “Grant Access” button next to its name:

Package ID Specifications

Package ID specifications

Subcommands of Cargo frequently need to refer to a particular package within a dependency graph for various operations like updating, cleaning, building, etc. To solve this problem, Cargo supports Package ID Specifications. A specification is a string which is used to uniquely refer to one package within a graph of packages.

Specification grammar

The formal grammar for a Package Id Specification is:

pkgid := pkgname
       | [ proto "://" ] hostname-and-path [ "#" ( pkgname | semver ) ]
pkgname := name [ ":" semver ]

proto := "http" | "git" | ...

Here, brackets indicate that the contents are optional.

Example specifications

These could all be references to a package foo version 1.2.3 from the registry at crates.io

Brevity of specifications

The goal of this is to enable both succinct and exhaustive syntaxes for referring to packages in a dependency graph. Ambiguous references may refer to one or more packages. Most commands generate an error if more than one package could be referred to with the same specification.

Source Replacement

This document is about replacing the crate index. You can read about overriding dependencies in the overriding dependencies section of this documentation.

A source is a provider that contains crates that may be included as dependencies for a package. Cargo supports the ability to replace one source with another to express strategies such as:

• Vendoring - custom sources can be defined which represent crates on the local filesystem. These sources are subsets of the source that they're replacing and can be checked into packages if necessary.

• Mirroring - sources can be replaced with an equivalent version which acts as a cache for crates.io itself.

Cargo has a core assumption about source replacement that the source code is exactly the same from both sources. Note that this also means that a replacement source is not allowed to have crates which are not present in the original source.

As a consequence, source replacement is not appropriate for situations such as patching a dependency or a private registry. Cargo supports patching dependencies through the usage of the [replace] key, and private registry support is described in Registries.

Configuration

Configuration of replacement sources is done through .cargo/config and the full set of available keys are:

# The `source` table is where all keys related to source-replacement
# are stored.
[source]

# Under the `source` table are a number of other tables whose keys are a
# name for the relevant source. For example this section defines a new
# source, called `my-vendor-source`, which comes from a directory
# located at `vendor` relative to the directory containing this `.cargo/config`
# file
[source.my-vendor-source]
directory = "vendor"

# The crates.io default source for crates is available under the name
# "crates-io", and here we use the `replace-with` key to indicate that it's
# replaced with our source above.
[source.crates-io]
replace-with = "my-vendor-source"

# Each source has its own table where the key is the name of the source
[source.the-source-name]

# Indicate that `the-source-name` will be replaced with `another-source`,
# defined elsewhere
replace-with = "another-source"

# Several kinds of sources can be specified (described in more detail below):
registry = "https://example.com/path/to/index"
local-registry = "path/to/registry"
directory = "path/to/vendor"

# Git sources can optionally specify a branch/tag/rev as well
git = "https://example.com/path/to/repo"
# branch = "master"
# tag = "v1.0.1"
# rev = "313f44e8"

Registry Sources

A "registry source" is one that is the same as crates.io itself. That is, it has an index served in a git repository which matches the format of the crates.io index. That repository then has configuration indicating where to download crates from.

Currently there is not an already-available project for setting up a mirror of crates.io. Stay tuned though!

Local Registry Sources

A "local registry source" is intended to be a subset of another registry source, but available on the local filesystem (aka vendoring). Local registries are downloaded ahead of time, typically sync'd with a Cargo.lock, and are made up of a set of *.crate files and an index like the normal registry is.

The primary way to manage and create local registry sources is through the cargo-local-registry subcommand, available on crates.io and can be installed with cargo install cargo-local-registry.

Local registries are contained within one directory and contain a number of *.crate files downloaded from crates.io as well as an index directory with the same format as the crates.io-index project (populated with just entries for the crates that are present).

Directory Sources

A "directory source" is similar to a local registry source where it contains a number of crates available on the local filesystem, suitable for vendoring dependencies. Directory sources are primarily managed the cargo vendor subcommand.

Directory sources are distinct from local registries though in that they contain the unpacked version of *.crate files, making it more suitable in some situations to check everything into source control. A directory source is just a directory containing a number of other directories which contain the source code for crates (the unpacked version of *.crate files). Currently no restriction is placed on the name of each directory.

Each crate in a directory source also has an associated metadata file indicating the checksum of each file in the crate to protect against accidental modifications.

External tools

One of the goals of Cargo is simple integration with third-party tools, like IDEs and other build systems. To make integration easier, Cargo has several facilities:

• a cargo metadata command, which outputs package structure and dependencies information in JSON,

• a --message-format flag, which outputs information about a particular build, and

• support for custom subcommands.

Information about package structure

You can use cargo metadata command to get information about package structure and dependencies. The output of the command looks like this:

{
  // Integer version number of the format.
  "version": integer,

  // List of packages for this workspace, including dependencies.
  "packages": [
    {
      // Opaque package identifier.
      "id": PackageId,

      "name": string,

      "version": string,

      "source": SourceId,

      // A list of declared dependencies, see `resolve` field for actual dependencies.
      "dependencies": [ Dependency ],

      "targets: [ Target ],

      // Path to Cargo.toml
      "manifest_path": string,
    }
  ],

  "workspace_members": [ PackageId ],

  // Dependencies graph.
  "resolve": {
     "nodes": [
       {
         "id": PackageId,
         "dependencies": [ PackageId ]
       }
     ]
  }
}

The format is stable and versioned. When calling cargo metadata, you should pass --format-version flag explicitly to avoid forward incompatibility hazard.

If you are using Rust, there is cargo_metadata crate.

Information about build

When passing --message-format=json, Cargo will output the following information during the build:

• compiler errors and warnings,

• produced artifacts,

• results of the build scripts (for example, native dependencies).

The output goes to stdout in the JSON object per line format. The reason field distinguishes different kinds of messages.

Information about dependencies in the Makefile-compatible format is stored in the .d files alongside the artifacts.

Custom subcommands

Cargo is designed to be extensible with new subcommands without having to modify Cargo itself. This is achieved by translating a cargo invocation of the form cargo (?<command>[^ ]+) into an invocation of an external tool cargo-${command}. The external tool must be present in one of the user's $PATH directories.

When Cargo invokes a custom subcommand, the first argument to the subcommand will be the filename of the custom subcommand, as usual. The second argument will be the subcommand name itself. For example, the second argument would be ${command} when invoking cargo-${command}. Any additional arguments on the command line will be forwarded unchanged.

Cargo can also display the help output of a custom subcommand with cargo help ${command}. Cargo assumes that the subcommand will print a help message if its third argument is --help. So, cargo help ${command} would invoke cargo-${command} ${command} --help.

Custom subcommands may use the CARGO environment variable to call back to Cargo. Alternatively, it can link to cargo crate as a library, but this approach has drawbacks:

• Cargo as a library is unstable: the API may change without deprecation

• versions of the linked Cargo library may be different from the Cargo binary

Registries

Cargo installs crates and fetches dependencies from a "registry". The default registry is crates.io. A registry contains an "index" which contains a searchable list of available crates. A registry may also provide a web API to support publishing new crates directly from Cargo.

Note: If you are interested in mirroring or vendoring an existing registry, take a look at Source Replacement.

Using an Alternate Registry

To use a registry other than crates.io, the name and index URL of the registry must be added to a .cargo/config file. The registries table has a key for each registry, for example:

[registries]
my-registry = { index = "https://my-intranet:8080/git/index" }

The index key should be a URL to a git repository with the registry's index. A crate can then depend on a crate from another registry by specifying the registry key and a value of the registry's name in that dependency's entry in Cargo.toml:

# Sample Cargo.toml
[package]
name = "my-project"
version = "0.1.0"

[dependencies]
other-crate = { version = "1.0", registry = "my-registry" }

As with most config values, the index may be specified with an environment variable instead of a config file. For example, setting the following environment variable will accomplish the same thing as defining a config file:

CARGO_REGISTRIES_MY_REGISTRY_INDEX=https://my-intranet:8080/git/index

Note: crates.io does not accept packages that depend on crates from other registries.

Publishing to an Alternate Registry

If the registry supports web API access, then packages can be published directly to the registry from Cargo. Several of Cargo's commands such as cargo publish take a --registry command-line flag to indicate which registry to use. For example, to publish the package in the current directory:

1. cargo login --registry=my-registry

   This only needs to be done once. You must enter the secret API token retrieved from the registry's website. Alternatively the token may be passed directly to the publish command with the --token command-line flag or an environment variable with the name of the registry such as CARGO_REGISTRIES_MY_REGISTRY_TOKEN.

2. cargo publish --registry=my-registry

Instead of always passing the --registry command-line option, the default registry may be set in .cargo/config with the registry.default key.

Setting the package.publish key in the Cargo.toml manifest restricts which registries the package is allowed to be published to. This is useful to prevent accidentally publishing a closed-source package to crates.io. The value may be a list of registry names, for example:

[package]
# ...
publish = ["my-registry"]

The publish value may also be false to restrict all publishing, which is the same as an empty list.

The authentication information saved by cargo login is stored in the credentials file in the Cargo home directory (default $HOME/.cargo). It has a separate table for each registry, for example:

[registries.my-registry]
token = "854DvwSlUwEHtIo3kWy6x7UCPKHfzCmy"

Running a Registry

A minimal registry can be implemented by having a git repository that contains an index, and a server that contains the compressed .crate files created by cargo package. Users won't be able to use Cargo to publish to it, but this may be sufficient for closed environments.

A full-featured registry that supports publishing will additionally need to have a web API service that conforms to the API used by Cargo. The web API is documented below.

At this time, there is no widely used software for running a custom registry. There is interest in documenting projects that implement registry support, or existing package caches that add support for Cargo.

Index Format

The following defines the format of the index. New features are occasionally added, which are only understood starting with the version of Cargo that introduced them. Older versions of Cargo may not be able to use packages that make use of new features. However, the format for older packages should not change, so older versions of Cargo should be able to use them.

The index is stored in a git repository so that Cargo can efficiently fetch incremental updates to the index. In the root of the repository is a file named config.json which contains JSON information used by Cargo for accessing the registry. This is an example of what the crates.io config file looks like:

{
    "dl": "https://crates.io/api/v1/crates",
    "api": "https://crates.io"
}

The keys are:

• dl: This is the URL for downloading crates listed in the index. The value may have the markers {crate} and {version} which are replaced with the name and version of the crate to download. If the markers are not present, then the value /{crate}/{version}/download is appended to the end.
• api: This is the base URL for the web API. This key is optional, but if it is not specified, commands such as cargo publish will not work. The web API is described below.

The download endpoint should send the .crate file for the requested package. Cargo supports https, http, and file URLs, HTTP redirects, HTTP1 and HTTP2. The exact specifics of TLS support depend on the platform that Cargo is running on, the version of Cargo, and how it was compiled.

The rest of the index repository contains one file for each package, where the filename is the name of the package in lowercase. Each version of the package has a separate line in the file. The files are organized in a tier of directories:

• Packages with 1 character names are placed in a directory named 1.
• Packages with 2 character names are placed in a directory named 2.
• Packages with 3 character names are placed in the directory 3/{first-character} where {first-character} is the first character of the package name.
• All other packages are stored in directories named {first-two}/{second-two} where the top directory is the first two characters of the package name, and the next subdirectory is the third and fourth characters of the package name. For example, cargo would be stored in a file named ca/rg/cargo.

Note: Although the index filenames are in lowercase, the fields that contain package names in Cargo.toml and the index JSON data are case-sensitive and may contain upper and lower case characters.

Registries may want to consider enforcing limitations on package names added to their index. Cargo itself allows names with any alphanumeric, -, or _ character. For example, crates.io imposes relatively strict limitations, such as requiring it to be a valid Rust identifier, only allowing ASCII characters, under a specific length, and rejects reserved names such as Windows special filenames like "nul".

Each line in a package file contains a JSON object that describes a published version of the package. The following is a pretty-printed example with comments explaining the format of the entry.

{
    // The name of the package.
    // This must only contain alphanumeric, `-`, or `_` characters.
    "name": "foo",
    // The version of the package this row is describing.
    // This must be a valid version number according to the Semantic
    // Versioning 2.0.0 spec at https://semver.org/.
    "vers": "0.1.0",
    // Array of direct dependencies of the package.
    "deps": [
        {
            // Name of the dependency.
            // If the dependency is renamed from the original package name,
            // this is the new name. The original package name is stored in
            // the `package` field.
            "name": "rand",
            // The semver requirement for this dependency.
            // This must be a valid version requirement defined at
            // https://github.com/steveklabnik/semver#requirements.
            "req": "^0.6",
            // Array of features (as strings) enabled for this dependency.
            "features": ["i128_support"],
            // Boolean of whether or not this is an optional dependency.
            "optional": false,
            // Boolean of whether or not default features are enabled.
            "default_features": true,
            // The target platform for the dependency.
            // null if not a target dependency.
            // Otherwise, a string such as "cfg(windows)".
            "target": null,
            // The dependency kind.
            // "dev", "build", or "normal".
            // Note: this is a required field, but a small number of entries
            // exist in the crates.io index with either a missing or null
            // `kind` field due to implementation bugs.
            "kind": "normal",
            // The URL of the index of the registry where this dependency is
            // from as a string. If not specified or null, it is assumed the
            // dependency is in the current registry.
            "registry": null,
            // If the dependency is renamed, this is a string of the actual
            // package name. If not specified or null, this dependency is not
            // renamed.
            "package": null,
        }
    ],
    // A SHA256 checksum of the `.crate` file.
    "cksum": "d867001db0e2b6e0496f9fac96930e2d42233ecd3ca0413e0753d4c7695d289c",
    // Set of features defined for the package.
    // Each feature maps to an array of features or dependencies it enables.
    "features": {
        "extras": ["rand/simd_support"]
    },
    // Boolean of whether or not this version has been yanked.
    "yanked": false,
    // The `links` string value from the package's manifest, or null if not
    // specified. This field is optional and defaults to null.
    "links": null
}

The JSON objects should not be modified after they are added except for the yanked field whose value may change at any time.

Web API

A registry may host a web API at the location defined in config.json to support any of the actions listed below.

Cargo includes the Authorization header for requests that require authentication. The header value is the API token. The server should respond with a 403 response code if the token is not valid. Users are expected to visit the registry's website to obtain a token, and Cargo can store the token using the cargo login command, or by passing the token on the command-line.

Responses use a 200 response code for both success and errors. Cargo looks at the JSON response to determine if there was success or failure. Failure responses have a JSON object with the following structure:

{
    // Array of errors to display to the user.
    "errors": [
        {
            // The error message as a string.
            "detail": "error message text"
        }
    ]
}

Servers may also respond with a 404 response code to indicate the requested resource is not found (for example, an unknown crate name). However, using a 200 response with an errors object allows a registry to provide a more detailed error message if desired.

For backwards compatibility, servers should ignore any unexpected query parameters or JSON fields. If a JSON field is missing, it should be assumed to be null. The endpoints are versioned with the v1 component of the path, and Cargo is responsible for handling backwards compatibility fallbacks should any be required in the future.

Cargo sets the following headers for all requests:

• Content-Type: application/json
• Accept: application/json
• User-Agent: The Cargo version such as cargo 1.32.0 (8610973aa 2019-01-02). This may be modified by the user in a configuration value. Added in 1.29.

Publish

• Endpoint: /api/v1/crates/new
• Method: PUT
• Authorization: Included

The publish endpoint is used to publish a new version of a crate. The server should validate the crate, make it available for download, and add it to the index.

The body of the data sent by Cargo is:

• 32-bit unsigned little-endian integer of the length of JSON data.
• Metadata of the package as a JSON object.
• 32-bit unsigned little-endian integer of the length of the .crate file.
• The .crate file.

The following is a commented example of the JSON object. Some notes of some restrictions imposed by crates.io are included only to illustrate some suggestions on types of validation that may be done, and should not be considered as an exhaustive list of restrictions crates.io imposes.

{
    // The name of the package.
    "name": "foo",
    // The version of the package being published.
    "vers": "0.1.0",
    // Array of direct dependencies of the package.
    "deps": [
        {
            // Name of the dependency.
            // If the dependency is renamed from the original package name,
            // this is the original name. The new package name is stored in
            // the `explicit_name_in_toml` field.
            "name": "rand",
            // The semver requirement for this dependency.
            "version_req": "^0.6",
            // Array of features (as strings) enabled for this dependency.
            "features": ["i128_support"],
            // Boolean of whether or not this is an optional dependency.
            "optional": false,
            // Boolean of whether or not default features are enabled.
            "default_features": true,
            // The target platform for the dependency.
            // null if not a target dependency.
            // Otherwise, a string such as "cfg(windows)".
            "target": null,
            // The dependency kind.
            // "dev", "build", or "normal".
            "kind": "normal",
            // The URL of the index of the registry where this dependency is
            // from as a string. If not specified or null, it is assumed the
            // dependency is in the current registry.
            "registry": null,
            // If the dependency is renamed, this is a string of the new
            // package name. If not specified or null, this dependency is not
            // renamed.
            "explicit_name_in_toml": null,
        }
    ],
    // Set of features defined for the package.
    // Each feature maps to an array of features or dependencies it enables.
    // Cargo does not impose limitations on feature names, but crates.io
    // requires alphanumeric ASCII, `_` or `-` characters.
    "features": {
        "extras": ["rand/simd_support"]
    },
    // List of strings of the authors.
    // May be empty. crates.io requires at least one entry.
    "authors": ["Alice <a@example.com>"],
    // Description field from the manifest.
    // May be null. crates.io requires at least some content.
    "description": null,
    // String of the URL to the website for this package's documentation.
    // May be null.
    "documentation": null,
    // String of the URL to the website for this package's home page.
    // May be null.
    "homepage": null,
    // String of the content of the README file.
    // May be null.
    "readme": null,
    // String of a relative path to a README file in the crate.
    // May be null.
    "readme_file": null,
    // Array of strings of keywords for the package.
    "keywords": [],
    // Array of strings of categories for the package.
    "categories": [],
    // String of the license for the package.
    // May be null. crates.io requires either `license` or `license_file` to be set.
    "license": null,
    // String of a relative path to a license file in the crate.
    // May be null.
    "license_file": null,
    // String of the URL to the website for the source repository of this package.
    // May be null.
    "repository": null,
    // Optional object of "status" badges. Each value is an object of
    // arbitrary string to string mappings.
    // crates.io has special interpretation of the format of the badges.
    "badges": {
        "travis-ci": {
            "branch": "master",
            "repository": "rust-lang/cargo"
        }
    },
    // The `links` string value from the package's manifest, or null if not
    // specified. This field is optional and defaults to null.
    "links": null,
}

A successful response includes the JSON object:

{
    // Optional object of warnings to display to the user.
    "warnings": {
        // Array of strings of categories that are invalid and ignored.
        "invalid_categories": [],
        // Array of strings of badge names that are invalid and ignored.
        "invalid_badges": [],
        // Array of strings of arbitrary warnings to display to the user.
        "other": []
    }
}

Yank

• Endpoint: /api/v1/crates/{crate_name}/{version}/yank
• Method: DELETE
• Authorization: Included

The yank endpoint will set the yank field of the given version of a crate to true in the index.

A successful response includes the JSON object:

{
    // Indicates the delete succeeded, always true.
    "ok": true,
}

Unyank

• Endpoint: /api/v1/crates/{crate_name}/{version}/unyank
• Method: PUT
• Authorization: Included

The unyank endpoint will set the yank field of the given version of a crate to false in the index.

A successful response includes the JSON object:

{
    // Indicates the delete succeeded, always true.
    "ok": true,
}

Owners

Cargo does not have an inherent notion of users and owners, but it does provide the owner command to assist managing who has authorization to control a crate. It is up to the registry to decide exactly how users and owners are handled. See the publishing documentation for a description of how crates.io handles owners via GitHub users and teams.

Owners: List

• Endpoint: /api/v1/crates/{crate_name}/owners
• Method: GET
• Authorization: Included

The owners endpoint returns a list of owners of the crate.

A successful response includes the JSON object:

{
    // Array of owners of the crate.
    "users": [
        {
            // Unique unsigned 32-bit integer of the owner.
            "id": 70,
            // The unique username of the owner.
            "login": "github:rust-lang:core",
            // Name of the owner.
            // This is optional and may be null.
            "name": "Core",
        }
    ]
}

Owners: Add

• Endpoint: /api/v1/crates/{crate_name}/owners
• Method: PUT
• Authorization: Included

A PUT request will send a request to the registry to add a new owner to a crate. It is up to the registry how to handle the request. For example, crates.io sends an invite to the user that they must accept before being added.

The request should include the following JSON object:

{
    // Array of `login` strings of owners to add.
    "users": ["login_name"]
}

A successful response includes the JSON object:

{
    // Indicates the add succeeded, always true.
    "ok": true,
    // A string to be displayed to the user.
    "msg": "user ehuss has been invited to be an owner of crate cargo"
}

Owners: Remove

• Endpoint: /api/v1/crates/{crate_name}/owners
• Method: DELETE
• Authorization: Included

A DELETE request will remove an owner from a crate. The request should include the following JSON object:

{
    // Array of `login` strings of owners to remove.
    "users": ["login_name"]
}

A successful response includes the JSON object:

{
    // Indicates the remove succeeded, always true.
    "ok": true
}

Search

• Endpoint: /api/v1/crates
• Method: GET
• Query Parameters:
  • q: The search query string.
  • per_page: Number of results, default 10, max 100.

The search request will perform a search for crates, using criteria defined on the server.

A successful response includes the JSON object:

{
    // Array of results.
    "crates": [
        {
            // Name of the crate.
            "name": "rand",
            // The highest version available.
            "max_version": "0.6.1",
            // Textual description of the crate.
            "description": "Random number generators and other randomness functionality.\n",
        }
    ],
    "meta": {
        // Total number of results available on the server.
        "total": 119
    }
}

Login

• Endpoint: /me

The "login" endpoint is not an actual API request. It exists solely for the cargo login command to display a URL to instruct a user to visit in a web browser to log in and retrieve an API token.

Unstable Features

Experimental Cargo features are only available on the nightly channel. You typically use one of the -Z flags to enable them. Run cargo -Z help to see a list of flags available.

-Z unstable-options is a generic flag for enabling other unstable command-line flags. Options requiring this will be called out below.

Some unstable features will require you to specify the cargo-features key in Cargo.toml.

no-index-update

• Original Issue: #3479

The -Z no-index-update flag ensures that Cargo does not attempt to update the registry index. This is intended for tools such as Crater that issue many Cargo commands, and you want to avoid the network latency for updating the index each time.

avoid-dev-deps

• Original Issue: #4988
• Stabilization Issue: #5133

When running commands such as cargo install or cargo build, Cargo currently requires dev-dependencies to be downloaded, even if they are not used. The -Z avoid-dev-deps flag allows Cargo to avoid downloading dev-dependencies if they are not needed. The Cargo.lock file will not be generated if dev-dependencies are skipped.

minimal-versions

• Original Issue: #4100
• Tracking Issue: #5657

When a Cargo.lock file is generated, the -Z minimal-versions flag will resolve the dependencies to the minimum semver version that will satisfy the requirements (instead of the greatest version).

The intended use-case of this flag is to check, during continuous integration, that the versions specified in Cargo.toml are a correct reflection of the minimum versions that you are actually using. That is, if Cargo.toml says foo = "1.0.0" that you don't accidentally depend on features added only in foo 1.5.0.

out-dir

• Original Issue: #4875
• Tracking Issue: #6790

This feature allows you to specify the directory where artifacts will be copied to after they are built. Typically artifacts are only written to the target/release or target/debug directories. However, determining the exact filename can be tricky since you need to parse JSON output. The --out-dir flag makes it easier to predictably access the artifacts. Note that the artifacts are copied, so the originals are still in the target directory. Example:

cargo +nightly build --out-dir=out -Z unstable-options

Profile Overrides

• Tracking Issue: rust-lang/rust#48683
• RFC: #2282

Profiles can be overridden for specific packages and custom build scripts. The general format looks like this:

cargo-features = ["profile-overrides"]

[package]
...

[profile.dev]
opt-level = 0
debug = true

# the `image` crate will be compiled with -Copt-level=3
[profile.dev.overrides.image]
opt-level = 3

# All dependencies (but not this crate itself or any workspace member)
# will be compiled with -Copt-level=2 . This includes build dependencies.
[profile.dev.overrides."*"]
opt-level = 2

# Build scripts or proc-macros and their dependencies will be compiled with
# `-Copt-level=3`. By default, they use the same rules as the rest of the
# profile.
[profile.dev.build-override]
opt-level = 3

Overrides can only be specified for dev and release profiles.

Config Profiles

• Tracking Issue: rust-lang/rust#48683
• RFC: #2282

Profiles can be specified in .cargo/config files. The -Z config-profile command-line flag is required to use this feature. The format is the same as in a Cargo.toml manifest. If found in multiple config files, settings will be merged using the regular config hierarchy. Config settings take precedence over manifest settings.

[profile.dev]
opt-level = 3

cargo +nightly build -Z config-profile

Namespaced features

• Original issue: #1286
• Tracking Issue: #5565

Currently, it is not possible to have a feature and a dependency with the same name in the manifest. If you set namespaced-features to true, the namespaces for features and dependencies are separated. The effect of this is that, in the feature requirements, dependencies have to be prefixed with crate:. Like this:

[package]
namespaced-features = true

[features]
bar = ["crate:baz", "foo"]
foo = []

[dependencies]
baz = { version = "0.1", optional = true }

To prevent unnecessary boilerplate from having to explicitly declare features for each optional dependency, implicit features get created for any optional dependencies where a feature of the same name is not defined. However, if a feature of the same name as a dependency is defined, that feature must include the dependency as a requirement, as foo = ["crate:foo"].

Build-plan

• Tracking Issue: #5579

The --build-plan argument for the build command will output JSON with information about which commands would be run without actually executing anything. This can be useful when integrating with another build tool. Example:

cargo +nightly build --build-plan -Z unstable-options

Metabuild

• Tracking Issue: rust-lang/rust#49803
• RFC: #2196

Metabuild is a feature to have declarative build scripts. Instead of writing a build.rs script, you specify a list of build dependencies in the metabuild key in Cargo.toml. A build script is automatically generated that runs each build dependency in order. Metabuild packages can then read metadata from Cargo.toml to specify their behavior.

Include cargo-features at the top of Cargo.toml, a metabuild key in the package, list the dependencies in build-dependencies, and add any metadata that the metabuild packages require under package.metadata. Example:

cargo-features = ["metabuild"]

[package]
name = "mypackage"
version = "0.0.1"
metabuild = ["foo", "bar"]

[build-dependencies]
foo = "1.0"
bar = "1.0"

[package.metadata.foo]
extra-info = "qwerty"

Metabuild packages should have a public function called metabuild that performs the same actions as a regular build.rs script would perform.

install-upgrade

• Tracking Issue: #6797

The install-upgrade feature changes the behavior of cargo install so that it will reinstall a package if it is not "up-to-date". If it is "up-to-date", it will do nothing and exit with success instead of failing. Example:

cargo +nightly install foo -Z install-upgrade

Cargo tracks some information to determine if a package is "up-to-date", including:

• The package version and source.
• The set of binary names installed.
• The chosen features.
• The release mode (--debug).
• The target (--target).

If any of these values change, then Cargo will reinstall the package.

Installation will still fail if a different package installs a binary of the same name. --force may be used to unconditionally reinstall the package.

Installing with --path will always build and install, unless there are conflicting binaries from another package.

Additionally, a new flag --no-track is available to prevent cargo install from writing tracking information in $CARGO_HOME about which packages are installed.

public-dependency

• Tracking Issue: #44663

The 'public-dependency' feature allows marking dependencies as 'public' or 'private'. When this feature is enabled, additional information is passed to rustc to allow the 'exported_private_dependencies' lint to function properly.

This requires the appropriate key to be set in cargo-features:

cargo-features = ["public-dependency"]

[dependencies]
my_dep = { version = "1.2.3", public = true }
private_dep = "2.0.0" # Will be 'private' by default

cache-messages

• Tracking Issue: #6986

The cache-messages feature causes Cargo to cache the messages generated by the compiler. This is primarily useful if a crate compiles successfully with warnings. Previously, re-running Cargo would not display any output. With the cache-messages feature, it will quickly redisplay the previous warnings.

cargo +nightly check -Z cache-messages

This works with any command that runs the compiler (build, check, test, etc.).

This also changes the way Cargo interacts with the compiler, helping to prevent interleaved messages when multiple crates attempt to display a message at the same time.

build-std

• Tracking Repository: https://github.com/rust-lang/wg-cargo-std-aware

The build-std feature enables Cargo to compile the standard library itself as part of a crate graph compilation. This feature has also historically been known as "std-aware Cargo". This feature is still in very early stages of development, and is also a possible massive feature addition to Cargo. This is a very large feature to document, even in the minimal form that it exists in today, so if you're curious to stay up to date you'll want to follow the tracking repository and its set of issues.

The functionality implemented today is behind a flag called -Z build-std. This flag indicates that Cargo should compile the standard library from source code using the same profile as the main build itself. Note that for this to work you need to have the source code for the standard library available, and at this time the only supported method of doing so is to add the rust-src rust rustup component:

$ rustup component add rust-src --toolchain nightly

It is also required today that the -Z build-std flag is combined with the --target flag. Note that you're not forced to do a cross compilation, you're just forced to pass --target in one form or another.

Usage looks like:

$ cargo new foo
$ cd foo
$ cargo +nightly run -Z build-std --target x86_64-unknown-linux-gnu
   Compiling core v0.0.0 (...)
   ...
   Compiling foo v0.1.0 (...)
    Finished dev [unoptimized + debuginfo] target(s) in 21.00s
     Running `target/x86_64-unknown-linux-gnu/debug/foo`
Hello, world!

Here we recompiled the standard library in debug mode with debug assertions (like src/main.rs is compiled) and everything was linked together at the end.

Using -Z build-std will implicitly compile the stable crates core, std, alloc, and proc_macro. If you're using cargo test it will also compile the test crate. If you're working with an environment which does not support some of these crates, then you can pass an argument to -Zbuild-std as well:

$ cargo +nightly build -Z build-std=core,alloc

The value here is a comma-separated list of standard library crates to build.

Requirements

As a summary, a list of requirements today to use -Z build-std are:

• You must install libstd's source code through rustup component add rust-src
• You must pass --target
• You must use both a nightly Cargo and a nightly rustc
• The -Z build-std flag must be passed to all cargo invocations.

Reporting bugs and helping out

The -Z build-std feature is in the very early stages of development! This feature for Cargo has an extremely long history and is very large in scope, and this is just the beginning. If you'd like to report bugs please either report them to:

• Cargo - https://github.com/rust-lang/cargo/issues/new - for implementation bugs
• The tracking repository - https://github.com/rust-lang/wg-cargo-std-aware/issues/new - for larger design questions.

Also if you'd like to see a feature that's not yet implemented and/or if something doesn't quite work the way you'd like it to, feel free to check out the issue tracker of the tracking repository, and if it's not there please file a new issue!

timings

The timings feature gives some information about how long each compilation takes, and tracks concurrency information over time.

cargo +nightly build -Z timings

The -Ztimings flag can optionally take a comma-separated list of the following values:

• html — Saves a file called cargo-timing.html to the current directory with a report of the compilation. Files are also saved with a timestamp in the filename if you want to look at older runs.
• info — Displays a message to stdout after each compilation finishes with how long it took.
• json — Emits some JSON information about timing information.

The default if none are specified is html,info.

Reading the graphs

There are two graphs in the output. The "unit" graph shows the duration of each unit over time. A "unit" is a single compiler invocation. There are lines that show which additional units are "unlocked" when a unit finishes. That is, it shows the new units that are now allowed to run because their dependencies are all finished. Hover the mouse over a unit to highlight the lines. This can help visualize the critical path of dependencies. This may change between runs because the units may finish in different orders.

The "codegen" times are highlighted in a lavender color. In some cases, build pipelining allows units to start when their dependencies are performing code generation. This information is not always displayed (for example, binary units do not show when code generation starts).

The "custom build" units are build.rs scripts, which when run are highlighted in orange.

The second graph shows Cargo's concurrency over time. The three lines are:

• "Waiting" (red) — This is the number of units waiting for a CPU slot to open.
• "Inactive" (blue) — This is the number of units that are waiting for their dependencies to finish.
• "Active" (green) — This is the number of units currently running.

Note: This does not show the concurrency in the compiler itself. rustc coordinates with Cargo via the "job server" to stay within the concurrency limit. This currently mostly applies to the code generation phase.

Tips for addressing compile times:

• Look for slow dependencies.
  • Check if they have features that you may wish to consider disabling.
  • Consider trying to remove the dependency completely.
• Look for a crate being built multiple times with different versions. Try to remove the older versions from the dependency graph.
• Split large crates into smaller pieces.
• If there are a large number of crates bottlenecked on a single crate, focus your attention on improving that one crate to improve parallelism.

cargo

NAME

Build Commands

cargo bench

NAME

cargo bench -- foo --exact

cargo bench -- --nocapture

cargo build

NAME

cargo check

NAME

cargo clean

NAME

cargo doc

NAME

cargo fetch

NAME

cargo fix

NAME

cargo fix --edition

cargo fix --edition --target x86_64-pc-windows-gnu

cargo fix --edition --no-default-features --features foo

cargo run

NAME

cargo rustc

NAME

cargo rustdoc

NAME

cargo test

NAME

cargo test foo -- --test-threads 3

cargo test -- --nocapture

cargo test -j 2 -- --test-threads=2

Manifest Commands

cargo generate-lockfile

NAME

cargo locate-project

NAME

cargo metadata

NAME

{
    /* Array of all packages in the workspace.
       It also includes all feature-enabled dependencies unless --no-deps is used.
    */
    "packages": [
        {
            /* The name of the package. */
            "name": "my-package",
            /* The version of the package. */
            "version": "0.1.0",
            /* The Package ID, a unique identifier for referring to the package. */
            "id": "my-package 0.1.0 (path+file:///path/to/my-package)",
            /* The license value from the manifest, or null. */
            "license": "MIT/Apache-2.0",
            /* The license-file value from the manifest, or null. */
            "license_file": "LICENSE",
            /* The description value from the manifest, or null. */
            "description": "Package description.",
            /* The source ID of the package. This represents where
               a package is retrieved from.
               This is null for path dependencies and workspace members.
               For other dependencies, it is a string with the format:
               - "registry+URL" for registry-based dependencies.
                 Example: "registry+https://github.com/rust-lang/crates.io-index"
               - "git+URL" for git-based dependencies.
                 Example: "git+https://github.com/rust-lang/cargo?rev=5e85ba14aaa20f8133863373404cb0af69eeef2c#5e85ba14aaa20f8133863373404cb0af69eeef2c"
            */
            "source": null,
            /* Array of dependencies declared in the package's manifest. */
            "dependencies": [
                {
                    /* The name of the dependency. */
                    "name": "bitflags",
                    /* The source ID of the dependency. May be null, see
                       description for the package source.
                    */
                    "source": "registry+https://github.com/rust-lang/crates.io-index",
                    /* The version requirement for the dependency.
                       Dependencies without a version requirement have a value of "*".
                    */
                    "req": "^1.0",
                    /* The dependency kind.
                       "dev", "build", or null for a normal dependency.
                    */
                    "kind": null,
                    /* If the dependency is renamed, this is the new name for
                       the dependency as a string.  null if it is not renamed.
                    */
                    "rename": null,
                    /* Boolean of whether or not this is an optional dependency. */
                    "optional": false,
                    /* Boolean of whether or not default features are enabled. */
                    "uses_default_features": true,
                    /* Array of features enabled. */
                    "features": [],
                    /* The target platform for the dependency.
                       null if not a target dependency.
                    */
                    "target": "cfg(windows)",
                    /* A string of the URL of the registry this dependency is from.
                       If not specified or null, the dependency is from the default
                       registry (crates.io).
                    */
                    "registry": null
                }
            ],
            /* Array of Cargo targets. */
            "targets": [
                {
                    /* Array of target kinds.
                       - lib targets list the `crate-type` values from the
                         manifest such as "lib", "rlib", "dylib",
                         "proc-macro", etc. (default ["lib"])
                       - binary is ["bin"]
                       - example is ["example"]
                       - integration test is ["test"]
                       - benchmark is ["bench"]
                       - build script is ["custom-build"]
                    */
                    "kind": [
                        "bin"
                    ],
                    /* Array of crate types.
                       - lib and example libraries list the `crate-type` values
                         from the manifest such as "lib", "rlib", "dylib",
                         "proc-macro", etc. (default ["lib"])
                       - all other target kinds are ["bin"]
                    */
                    "crate_types": [
                        "bin"
                    ],
                    /* The name of the target. */
                    "name": "my-package",
                    /* Absolute path to the root source file of the target. */
                    "src_path": "/path/to/my-package/src/main.rs",
                    /* The Rust edition of the target.
                       Defaults to the package edition.
                    */
                    "edition": "2018",
                    /* Array of required features.
                       This property is not included if no required features are set.
                    */
                    "required-features": ["feat1"],
                    /* Whether or not this target has doc tests enabled, and
                       the target is compatible with doc testing.
                    */
                    "doctest": false
                }
            ],
            /* Set of features defined for the package.
               Each feature maps to an array of features or dependencies it
               enables.
            */
            "features": {
                "default": [
                    "feat1"
                ],
                "feat1": [],
                "feat2": []
            },
            /* Absolute path to this package's manifest. */
            "manifest_path": "/path/to/my-package/Cargo.toml",
            /* Package metadata.
               This is null if no metadata is specified.
            */
            "metadata": {
                "docs": {
                    "rs": {
                        "all-features": true
                    }
                }
            },
            /* Array of authors from the manifest.
               Empty array if no authors specified.
            */
            "authors": [
                "Jane Doe <user@example.com>"
            ],
            /* Array of categories from the manifest. */
            "categories": [
                "command-line-utilities"
            ],
            /* Array of keywords from the manifest. */
            "keywords": [
                "cli"
            ],
            /* The readme value from the manifest or null if not specified. */
            "readme": "README.md",
            /* The repository value from the manifest or null if not specified. */
            "repository": "https://github.com/rust-lang/cargo",
            /* The default edition of the package.
               Note that individual targets may have different editions.
            */
            "edition": "2018",
            /* Optional string that is the name of a native library the package
               is linking to.
            */
            "links": null,
        }
    ],
    /* Array of members of the workspace.
       Each entry is the Package ID for the package.
    */
    "workspace_members": [
        "my-package 0.1.0 (path+file:///path/to/my-package)",
    ],
    /* The resolved dependency graph, with the concrete versions and features
       selected. The set depends on the enabled features.
       This is null if --no-deps is specified.
    */
    "resolve": {
        /* Array of nodes within the dependency graph.
           Each node is a package.
        */
        "nodes": [
            {
                /* The Package ID of this node. */
                "id": "my-package 0.1.0 (path+file:///path/to/my-package)",
                /* The dependencies of this package, an array of Package IDs. */
                "dependencies": [
                    "bitflags 1.0.4 (registry+https://github.com/rust-lang/crates.io-index)"
                ],
                /* The dependencies of this package. This is an alternative to
                   "dependencies" which contains additional information. In
                   particular, this handles renamed dependencies.
                */
                "deps": [
                    {
                        /* The name of the dependency's library target.
                           If this is a renamed dependency, this is the new
                           name.
                        */
                        "name": "bitflags",
                        /* The Package ID of the dependency. */
                        "pkg": "bitflags 1.0.4 (registry+https://github.com/rust-lang/crates.io-index)"
                    }
                ],
                /* Array of features enabled on this package. */
                "features": [
                    "default"
                ]
            }
        ],
        /* The root package of the workspace.
           This is null if this is a virtual workspace. Otherwise it is
           the Package ID of the root package.
        */
        "root": "my-package 0.1.0 (path+file:///path/to/my-package)"
    },
    /* The absolute path to the build directory where Cargo places its output. */
    "target_directory": "/path/to/my-package/target",
    /* The version of the schema for this metadata structure.
       This will be changed if incompatible changes are ever made.
    */
    "version": 1,
    /* The absolute path to the root of the workspace. */
    "workspace_root": "/path/to/my-package"
}

cargo pkgid

NAME

cargo update

NAME