Cargo packages consist of targets which correspond to source files which can
be compiled into a crate. Packages can have library,
binary, example, test, and
benchmark targets. The list of targets can be configured in the
Cargo.toml manifest, often inferred automatically
by the directory layout of the source files.
See Configuring a target below for details on configuring the settings for a target.
The library target defines a "library" that can be used and linked by other
libraries and executables. The filename defaults to
src/lib.rs, and the name
of the library defaults to the name of the package. A package can have only
one library. The settings for the library can be customized in the
# Example of customizing the library in Cargo.toml. [lib] crate-type = ["cdylib"] bench = false
Binary targets are executables programs that can be run after being compiled.
The default binary filename is
src/main.rs, which defaults to the name of
the package. Additional binaries are stored in the
directory. The settings for each binary can be customized
[[bin]] tables in
Binaries can use the public API of the package's library. They are also linked
[dependencies] defined in
# Example of customizing binaries in Cargo.toml. [[bin]] name = "cool-tool" test = false bench = false [[bin]] name = "frobnicator" required-features = ["frobnicate"]
By default, examples are executable binaries (with a
main() function). You
can specify the
crate-type field to make an example
be compiled as a library:
[[example]] name = "foo" crate-type = ["staticlib"]
You can run individual executable examples with the
cargo run command with
--example <example-name> option. Library examples can be built with
cargo build with the
--example <example-name> option.
--example <example-name> option can be used to copy executable
binaries to a common location. Examples are compiled by
cargo test by
default to protect them from bit-rotting. Set the
true if you have
#[test] functions in the
example that you want to run with
There are two styles of tests within a Cargo project:
- Unit tests which are functions marked with the
#[test]attribute located within your library or binaries (or any target enabled with the
testfield). These tests have access to private APIs located within the target they are defined in.
- Integration tests which is a separate executable binary, also containing
#[test]functions, which is linked with the project's library and has access to its public API.
Tests are run with the
cargo test command. By default, Cargo and
use the libtest harness which is responsible for collecting functions
annotated with the
#[test] attribute and executing them in
parallel, reporting the success and failure of each test. See the
field if you want to use a different harness or test
If you want to share code among multiple integration tests, you can place it
in a separate module such as
tests/common/mod.rs and then put
in each test to import it.
Each integration test results in a separate executable binary, and
cargo test will run them serially. In some cases this can be inefficient, as it
can take longer to compile, and may not make full use of multiple CPUs when
running the tests. If you have a lot of integration tests, you may want to
consider creating a single integration test, and split the tests into multiple
modules. The libtest harness will automatically find all of the
annotated functions and run them in parallel. You can pass module names to
cargo test to only run the tests within that module.
Benchmarks provide a way to test the performance of your code using the
cargo bench command. They follow the same structure as tests,
with each benchmark function annotated with the
Similarly to tests:
- Benchmarks are placed in the
- Benchmark functions defined in libraries and binaries have access to the
private API within the target they are defined in. Benchmarks in the
benchesdirectory may use the public API.
benchfield can be used to define which targets are benchmarked by default.
harnessfield can be used to disable the built-in harness.
#[bench]attribute is currently unstable and only available on the nightly channel. There are some packages available on crates.io that may help with running benchmarks on the stable channel, such as Criterion.
All of the
Cargo.toml support similar configuration for specifying how a
target should be built. The double-bracket sections like
array-of-table of TOML,
which means you can write more than one
[[bin]] section to make several
executables in your crate. You can only specify one library, so
[lib] is a
normal TOML table.
The following is an overview of the TOML settings for each target, with each field described in detail below.
[lib] name = "foo" # The name of the target. path = "src/lib.rs" # The source file of the target. test = true # Is tested by default. doctest = true # Documentation examples are tested by default. bench = true # Is benchmarked by default. doc = true # Is documented by default. plugin = false # Used as a compiler plugin (deprecated). proc-macro = false # Set to `true` for a proc-macro library. harness = true # Use libtest harness. edition = "2015" # The edition of the target. crate-type = ["lib"] # The crate types to generate. required-features =  # Features required to build this target (N/A for lib).
name field specifies the name of the target, which corresponds to the
filename of the artifact that will be generated. For a library, this is the
crate name that dependencies will use to reference it.
[lib] and the default binary (
src/main.rs), this defaults to the
name of the package, with any dashes replaced with underscores. For other
auto discovered targets, it defaults to the
directory or file name.
This is required for all targets except
path field specifies where the source for the crate is located, relative
If not specified, the inferred path is used based on the target name.
test field indicates whether or not the target is tested by default by
cargo test. The default is
true for lib, bins, and tests.
Note: Examples are built by
cargo testby default to ensure they continue to compile, but they are not tested by default. Setting
test = truefor an example will also build it as a test and run any
#[test]functions defined in the example.
doctest field indicates whether or not documentation examples are
tested by default by
cargo test. This is only relevant for libraries, it
has no effect on other sections. The default is
true for the library.
bench field indicates whether or not the target is benchmarked by
cargo bench. The default is
true for lib, bins, and
doc field indicates whether or not the target is included in the
documentation generated by
cargo doc by default. The default is
libraries and binaries.
Note: The binary will be skipped if its name is the same as the lib target.
This field is used for
rustc plugins, which are being deprecated.
harness field indicates that the
--test flag will be passed to
rustc which will automatically include the libtest library which is the
driver for collecting and running tests marked with the
attribute or benchmarks with the
#[bench] attribute. The
true for all targets.
If set to
false, then you are responsible for defining a
to run tests and benchmarks.
Tests have the
cfg(test) conditional expression enabled whether
or not the harness is enabled.
edition field defines the Rust edition the target will use. If not
specified, it defaults to the
edition field for the
[package]. This field should usually not be set, and is only intended for
advanced scenarios such as incrementally transitioning a large package to a
crate-type field defines the crate types that will be generated by the
target. It is an array of strings, allowing you to specify multiple crate
types for a single target. This can only be specified for libraries and
examples. Binaries, tests, and benchmarks are always the "bin" crate type. The
The available options are
proc-macro. You can read more about the different crate
types in the Rust Reference Manual.
required-features field specifies which features the target needs in
order to be built. If any of the required features are not enabled, the
target will be skipped. This is only relevant for the
[[example]] sections, it has no effect on
[features] # ... postgres =  sqlite =  tools =  [[bin]] name = "my-pg-tool" required-features = ["postgres", "tools"]
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
[[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
false in the
[package] section will
disable auto-discovery of the corresponding target type.
[package] # ... autobins = false autoexamples = false autotests = false autobenches = false
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
falseif at least one target is manually defined in
Cargo.toml. Beginning with the 2018 edition, the default is always