Hello Beo

1. Create and build a new project​

To create a new project, we'll use the new command. Our project:

snarkvm new foo

This will create foo directory and the files with the basic structure of the project:

• README.md having the skeleton of a README with instructions on how to compile.

• main.beo the main file of the source code.

• program.json containing the identification of the project in JSON format. Particularly, a dev address and its private key for the program.

The main.beo file should have contents like this:

// The 'foo.beo' program.
program foo.beo;

function hello:
    input r0 as u32.public;
    input r1 as u32.private;
    add r0 r1 into r2;
    output r2 as u32.private;

You can run a program with the snarkvm run command, followed by the function name you want to run and its input parameters:

snarkvm run hello 2u32 3u32

You will see output like this:

 • Loaded universal setup (in 1478 ms)

⛓  Constraints

 •  'foo.beo/hello' - 33 constraints (called 1 time)

➡️  Output

 • 5u32

✅ Finished 'foo.beo/hello' (in "/Users/collin/code/snarkVM/foo")

As you can see, the output has the 5u32 value, representing the sum of the inputs.

2. Executing a program​

You can execute a program with the snarkvm execute command, followed by the function name you want to execute and its input parameters:

snarkvm execute hello 2u32 3u32

When the execution is finished, you should see the following output:

 • Loaded universal setup (in 1478 ms)

⛓  Constraints

 •  'foo.beo/hello' - 33 constraints (called 1 time)

➡️  Output

 • 5u32
 
  {"type":"execute","id":"at1 ... (transaction object truncated for brevity)

✅ Executed 'foo.beo/hello' (in "/Users/collin/code/snarkVM/foo")

As you can see, the output has the 5u32 value, representing the sum of the inputs.

A "universal setup" is loaded into your environment. You can read more about this in the Marlin paper.

Once the universal setup is ready, every function in your main.beo file is built, generating this in the output folder:

• hello.prover the prover for the hello function.

• hello.verifier the verifier for the hello function.

• main.avm the bytecode of your beo program to be run by the VM.

As you can already guess, we have only one .avm file for the whole program, but a prover and verifier for every function.

3. Overview of a program​

Let's examine the foo program inside the main.beo file:

// The 'foo.beo' program.
program foo.beo;

function hello:
    input r0 as u32.public;
    input r1 as u32.private;
    add r0 r1 into r2;
    output r2 as u32.private;

First, we need to declare the program as the following:

program foo.beo;

Afterwards, we can start writing its functions (or other Beo structures such as structs, records, closures, as we will see later).

In the case of functions we have it very easy:

function [function_name]:

The functions are composed of three main parts:

• The input section

  Here we declare its input parameters:

      input r0 as u32.public;
      input r1 as u32.private;

  Everything in Beo instructions are declared/stored inside a register with a type (i8,field,bool, etc.) and a visibility option (public or private), registers are named as r0, r1, ..., rn.

  In this case we use r0 and r1 to store the inputs passed in sequential order to a program as u32 values, where we can store 32-bit unsigned integers to perform our sum operation.

• The instructions section

  The next section consists of the core of our function: here we call the Beo Instructions we need to make our program do what we want. For example, performing an addition operation:

      add r0 r1 into r2;

  Every beo instruction is followed by its input parameters with its specific types, and the result is stored in the register specified after into.

  You can find all the available Beo instruction opcodes here.

• The output section

  Similar to the input section, the output section does the same for the output of the program. It's the return of the function.

      output r2 as u32.private;

4. Types​

Beo uses a strongly-typed syntax. The language supports 16 primitive types, and allows users to define custom types.

The Beo primitive types include:

address
boolean
field
group
i8
i16
i32
i64
i128
u8
u16
u32
u64
u128
scalar

Users can define custom types using the struct or record keywords. We will explore these in the next few sections.

4.1 Registers​

Registers are the places where you store data to then be able to modify it.

4.2 Structs​

Structs are user-defined data structures. They are very much like traditional structs in conventional programming languages. You can store structs into registers, like with any other Beo data types.

For example, let's build a struct representing a fixed-size array of 3 elements. Add this at the bottom of the main.beo file:

struct array3:
    a0 as u32;
    a1 as u32;
    a2 as u32;

Now, just for example purposes, let's code a function that adds one to each element of a register with an array3 data type stored in it.

function sum_one_to_array3:
    input r0 as array3.private;
    add r0.a0 1u32 into r1;
    add r0.a1 1u32 into r2;
    add r0.a2 1u32 into r3;
    cast r1 r2 r3 into r4 as array3;
    output r4 as array3.private;

As you can see, we can input a struct into register r0 and access struct elements with the . syntax. We perform the add instruction on every element, storing the results in registers r1, r2 and r3 and, finally, we make use of the cast command to create a new array3 struct into r4.

Now, let's run it. In this case, the only new thing you need to know is that structs are passed to the cli in the following format:

"{a0: 1u32, a1: 2u32, a2: 3u32}"

Now we can execute the snarkvm run command. We will clean the project to pick up the new code:

snarkvm clean && snarkvm run sum_one_to_array3 "{a0: 0u32, a1: 1u32, a2: 2u32}"

And we get the new array3 element as output:

➡️  Output
 • {
  a0: 1u32,
  a1: 2u32,
  a2: 3u32
}
✅ Executed 'foo.beo/sum_one_to_array3' (in "[...]/foo")

4.3 Records​

A record is a fundamental data structure for encoding user assets and application state. Records are very similar to structs, but they have one required component:

record token:
    owner as address.private

The owner refers to the Beo address that owns the record.

Records are important because they represent the basic Beo structure to handle state in your application.

When running an Beo function, only registers that belong to the application address can be passed as input registers. Otherwise, an error is raised and the application doesn't run.

You can find your development application address inside the .env file:

{
    NETWORK=testnet
    PRIVATE_KEY=APrivateKey1zkpFsQNXJwdvjKs9bRsM91KcwJW1gW4CDtF3FJbgVBAvPds
}

4.4 Beo State​

In Beo, the state of an application is managed through records. An Beo account can create a transaction to consume a record and produce a new record in its place. Records in Beo are encrypted to the record owner address, ensuring that all records in Beo are fully private.

5. Your first Beo Program: Making a transfer​

Consider this program:

// The 'foo.beo' program.
program foo.beo;

record token:
    owner as address.private;
    amount as u64.private;

function mint:
    input r0 as u64.private;
    cast self.signer r0 into r1 as token.record;
    output r1 as token.record;

function transfer_amount:
    //  sender token record
    input r0 as token.record;
    // receiver address
    input r1 as address.private;
    // amount to transfer
    input r2 as u64.private;
    // final balance of sender
    sub r0.amount r2 into r3;
    // final balance of receiver
    add 0u64 r2 into r4;
    // sender token record after the transfer
    cast r0.owner r3 into r5 as token.record;
    // receiver token record after the transfer
    cast r1 r4 into r6 as token.record;
    // sender new token record
    output r5 as token.record;
    // receiver new token record
    output r6 as token.record;

First, we define our own record data type called token, that has the required parameter owner and a user-defined parameter called amount, representing the amount of tokens we have.

This transfer_amount function receives 3 input parameters (sender record, receiver record and amount) and stores them in 3 registers (r0, r1 and r2). After that, it computes the final balance for both of them and stores it in r3 and r4 (using sub and add instructions to compute the subtraction and addition respectively). With those final amounts, it creates the output records for sender and receiver, storing them in r5 and r6. Finally, both records are sent out of the function with the output instruction.

To run this function, the first parameter is the input record of the program. The format of this parameter is the same as for struct types:

{
  owner: beo1x5nz5u4j50w482t5xtqc3jdwly9s8saaxlgjz0wvmuzmxv2l5q9qmypx09.private,
  amount: 50u64.private
}

Where:

• owner: the public address of the program, as found in the PRIVATE_KEY of the .env file.

• other parameters: depending on the program itself (in this example, we used the parameter amount with the value 50).

Let's run the transfer_amount function (if you are following along, remember to use the address found in the program.json for the owner field):

snarkvm clean && snarkvm run transfer_amount "{
owner: beo1x5nz5u4j50w482t5xtqc3jdwly9s8saaxlgjz0wvmuzmxv2l5q9qmypx09.private,
amount: 50u64.private,
_nonce: 0group.public
}" beo1h3gu7fky36y8r7v2x9phc434fgf20g8qd7c7u45v269jfw6vmugqjegcvp 10u64

We get the following output records:

⛓  Constraints

 •  'foo.beo/transfer_amount' - 4,172 constraints (called 1 time)
 
➡️  Outputs
 • {
  owner: beo1x5nz5u4j50w482t5xtqc3jdwly9s8saaxlgjz0wvmuzmxv2l5q9qmypx09.private,
  amount: 40u64.private
  _nonce: 2293253577170800572742339369209137467208538700597121244293392265726446806023group.public
}
 • {
  owner: beo1h3gu7fky36y8r7v2x9phc434fgf20g8qd7c7u45v269jfw6vmugqjegcvp.private,
  amount: 10u64.private
  _nonce: 2323253577170856894742339369235137467208538700597121244293392765726742543235group.public
}
✅ Finished 'foo.beo/transfer_amount' (in "[...]/foo")

And that's it. You have transferred your first owner-defined tokens in Beo!

Note: the _nonce is not written in Beo instructions. The compiler outputs the _nonce in record outputs. The user needs to provide it as input when using a record.