Okay, so we've installed the compiler. Yay! But how do we use it?

The first step is to open up a command-line window, and navigate to the directory where we want to store our new project. Then we can create the skeleton of a new program with cargo new foo.

When we do this, cargo will create a new directory named foo and set up the skeletal program inside it.

The default is for cargo to create the skeleton of an executable program, but we can also tell it to set up a new library for us. All that takes is an additional command-line argument (bar is the name of the new directory that will be created, like foo): cargo new --lib bar.

When we look inside the newly created foo directory, we see a file called Cargo.toml and a sub-directory called src. There may also be a Git version control repository, which we will ignore for now.

The Cargo.toml file is where metadata about the program is stored. That includes the program's name, version number, and authors, but importantly it also has a section for dependencies. Editing the content of the [dependencies] section is how we tell Rust that our code should be linked to external libraries when it is compiled, which libraries and versions to use, and where to find them. External libraries are collections of source code that were packaged up in order to make them easy to use as components of other programs. By finding and linking good libraries, we can save the time and effort of writing our whole program ourselves. Instead, we can write only the part that nobody else has already done.

If a library that our program depends on is published on https://crates.io/, all we have to do to link it is add its linking code to the dependencies section. Let's say we want to use serde (a tool for turning Rust data into formats such as JSON and back) in our program. First, we find its linking code with: cargo search serde.

This will print out a list of matches that looks something like this:

     serde = "1.0.70"                          # A generic serialization/deserialization framework
     serde_json = "1.0.24"                 # A JSON serialization file format
     serde_derive_internals = "0.23.1"     # AST representation used by Serde derive macros. Unstable.
     serde_any = "0.5.0"                   # Dynamic serialization and deserialization with the format chosen at runtime
     serde_yaml = "0.7.5"                 # YAML support for Serde
     serde_bytes = "0.10.4"              # Optimized handling of `&[u8]` and `Vec<u8>` for Serde
     serde_traitobject = "0.1.0"       # Serializable trait objects.  This library enables the serialization of trait objects such…
     cargo-ssearch = "0.1.2"             # cargo-ssearch: cargo search on steroids
     serde_codegen_internals = "0.14.2"    # AST representation used by Serde codegen. Unstable.
     serde_millis = "0.1.1"                #     A serde wrapper that stores integer millisecond value for timestamps     and duration…
     ... and 458 crates more (use --limit N to see more)

The first one is the core serde library, and the linking code is the part of the line before the # symbol. All we have to do is copy and paste that into the dependencies section of Cargo.toml, and Rust will know that it should compile and link serde when it compiles our foo program. So, the dependencies section of Cargo.toml would look like this:

 [dependencies]
 serde = "1.0.70"

Depending on a library stored in the Git version control system, either locally or remotely, is also easy. The linking code is slightly different, but it looks like this:

    [dependencies]
    thing = { git = "https://github.com/example/thing" }

We tell Rust where to find the repository, and it knows how to check it out, compile it, and link it with our program. The repository location doesn't have to be a URL; it can be any repository location that the git command recognizes.

We can also link against other libraries stored on our own systems, of course. To do this, we just have to add an entry such as this to our Cargo.toml file:

     [dependencies]
     example = { path = "/path/to/example" }

The path can be absolute or relative. If it's relative, it's interpreted as being relative to the directory containing our Cargo.toml file.

When creating an executable program, cargo adds a file called main.rs to our project as it is created. For a newly created library, it instead adds lib.rs. In either case, that file is the entry point for the whole project.

Let's take a look at the boilerplate main.rs file:

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

Simple enough, right? Cargo's default program is a Rust version of the classic hello world program, which has been re-implemented countless times by new programmers in every conceivable programming language.

If we look at a new library's lib.rs file, things are a little more interesting:

     #[cfg(test)]
     mod tests {
         #[test]
         fn it_works() {
             assert_eq!(2 + 2, 4);
         }
     }

Instead of having a main function, which all executable programs need because they need a place to start, the library boilerplate includes a framework for automated tests and a single test that confirms that 2 + 2 = 4.