This action lets you easily cross-compile Rust projects using cross.
Here's a simplified example from the test and release workflow for
my tool ubi
:
jobs:
release:
name: Release - ${{ matrix.platform.os-name }}
strategy:
matrix:
platform:
- os-name: FreeBSD-x86_64
runs-on: ubuntu-20.04
target: x86_64-unknown-freebsd
skip_tests: true
- os-name: Linux-x86_64
runs-on: ubuntu-20.04
target: x86_64-unknown-linux-musl
- os-name: Linux-aarch64
runs-on: ubuntu-20.04
target: aarch64-unknown-linux-musl
- os-name: Linux-riscv64
runs-on: ubuntu-20.04
target: riscv64gc-unknown-linux-gnu
- os-name: Windows-x86_64
runs-on: windows-latest
target: x86_64-pc-windows-msvc
- os-name: macOS-x86_64
runs-on: macOS-latest
target: x86_64-apple-darwin
# more targets here ...
runs-on: ${{ matrix.platform.runs-on }}
steps:
- name: Checkout
uses: actions/checkout@v4
- name: Build binary
uses: houseabsolute/actions-rust-cross@v1
with:
command: ${{ matrix.platform.command }}
target: ${{ matrix.platform.target }}
args: "--locked --release"
strip: true
- name: Publish artifacts and release
uses: houseabsolute/actions-rust-release@v0
with:
executable-name: ubi
target: ${{ matrix.platform.target }}
Note that for Linux or BSD targets, you should always set the runs-on
key to a Linux x86-64
architecture runner.
If you only want to do native ARM compilation, for example using the ubuntu-latest-arm
runner,
then there's no need to use this action. However, if you want to compile for many platforms,
including Linux ARM, using this action will simplify your config. This action is only tested on
Ubuntu x86-64, Windows, and macOS runners.
This action takes the following parameters:
Key | Type | Required? | Description |
---|---|---|---|
command |
string (one of build , test , both (build and test), or bench ) |
no | The command(s) to run. The default is build . Running the test command will fail with *BSD targets and non-x86 Windows. |
target |
string | yes | The target triple to compile for. This should be one of the targets found by running rustup target list . |
working-directory |
string | no | The working directory in which to run the cargo or cross commands. Defaults to the current directory (. ). |
toolchain |
string (one of stable , beta , or nightly ) |
no | The Rust toolchain version to install. The default is stable . |
GITHUB_TOKEN |
string | no | Defaults to the value of ${{ github.token }} . |
args |
string | no | A string-separated list of arguments to be passed to cross build , like --release --locked . |
strip |
boolean (true or false ) |
no | If this is true, then the resulting binaries will be stripped if possible. This is only possible for binaries which weren't cross-compiled. |
cross-version |
string | no | This can be used to set the version of cross to use. If specified, it should be a specific cross release tag (like v0.2.3 ) or a git ref (commit hash, HEAD , etc.). If this is not set then the latest released version will always be used. If this is set to a git ref then the version corresponding to that ref will be installed. |
use-rust-cache |
boolean | no | Whether or not to use the Swatinem/rust-cache@v2 action. This defaults to true. |
rust-cache-parameters |
string (containing JSON) | no | This must be a string containing valid JSON. The JSON should be an object where the keys are the parameters for the Swatinem/rust-cache@v2 action. |
Under the hood, this action will compile your binaries with either cargo
or cross
, depending on
the host machine and target. For Linux builds, it will always use cross
except for builds
targeting an x86 architecture like x86_64
or i686
.
On Windows and macOS, it's possible to compile for all supported targets out of the box, so cross
will not be used on those platforms.
If it needs to install cross
, it will install the latest version by downloading a release using
my tool ubi
. This is much faster than using cargo
to
build cross
.
When compiling on Windows, it will do so in a Powershell environment, which can matter in some
corner cases, like compiling the openssl
crate with the vendored
feature.
When running cargo
on a Linux system, it will also include the output of running
lsb_release --short --description
in the cache key. This is important for crates that link against
system libraries. If those library versions change across OS versions (e.g. Ubuntu 20.04 to 22.04),
then the cache will be broken for these cases.
When running cross
, the hash of the cross
binary will be included in the cache key. This is done
because the Docker images that cross
uses can change when cross
is updated. We want to make sure
that we do not re-use the cache across changes when these images change.
Finally, it will run strip
to strip the binaries it builds if the strip
parameter is true. This
is only possible for builds that are not done via cross
. In addition, Windows builds for aarch64
cannot be stripped either.
By default, this action will use
the Swatinem/rust-cache@v2
action to cache compiled
dependencies for a crate. Note that per the documentation for the rust-cache
action, it has fairly
limited value for crates without a Cargo.lock
file. The key
parameter passed to this action will
always include the value of the target
input. If you specify a key
parameter in
rust-cache-parameters
, then the target
input will be appended to the value you specify.
In my testing, it seemed like in some cases restoring the cache would delete existing files in a
target
directory. This manifested with this sequence of actions:
- Run
actions-rust-cross
to compile a crate in a top-level directory. - Run
actions-rust-cross
to compile a crate in a subdirectory.
After step 2, the compiled binaries from step 1 were no longer present, sometimes. I'm not sure exactly what's going on here, but my recommendation is to structure your workflows so that this cannot affect you.
For example, if you have multiple crates, each of which builds a binary you want to release, then you can avoid this issue by structuring your workflow as follows:
- Run
actions-rust-cross
to compile crate A. - Run the release steps for crate A.
- Run
actions-rust-cross
to compile crate B. - Run the release steps for crate B.
When structured this way, it does not matter if the output of crate A is deleted in step 3.