Skip to main content

Intro to Anchor development

Summary​

  • Programs on Solana have instruction handlers that execute instruction logic.
  • Rust is the most common language for building Solana programs. The Anchor framework takes care of common grunt work - like reading data from incoming instructions, and checking the right accounts are provided - so you can focus on building your Solana program.

Lesson​

Solana's ability to run arbitrary executable code is part of what makes it so powerful. Solana programs, similar to "smart contracts" in other blockchain environments, are quite literally the backbone of the Solana ecosystem. And the collection of programs grows daily as developers and creators dream up and deploy new programs.

This lesson will give you a basic introduction to writing and deploying a Solana program using the Rust programming language and the Anchor framework.

What is Anchor?​

Anchor makes writing Solana programs easier, faster, and more secure, making it the "go-to" framework for Solana development. It makes it easier to organize and reason about your code, implements common security checks automatically, and removes a significant amount of boilerplate code that is otherwise associated with writing a Solana program.

Anchor program structure​

Anchor uses macros and traits to generate boilerplate Rust code for you. These provide a clear structure to your program so you can more easily reason about your code. The main high-level macros and attributes are:

  • declare_id - a macro for declaring the program’s onchain address
  • #[program] - an attribute macro used to denote the module containing the program’s instruction logic
  • Accounts - a trait applied to structs representing the list of accounts required for an instruction
  • #[account] - an attribute macro used to define custom account types for the program

Let's talk about each of them before putting all the pieces together.

Declare your program ID​

The declare_id macro is used to specify the onchain address of the program (i.e. the programId). When you build an Anchor program for the first time, the framework will generate a new keypair. This becomes the default keypair used to deploy the program unless specified otherwise. The corresponding public key should be used as the programId specified in the declare_id! macro.

declare_id!("Fg6PaFpoGXkYsidMpWTK6W2BeZ7FEfcYkg476zPFsLnS");

Define instruction logic​

The #[program] attribute macro defines the module containing all of your program's instructions. This is where you implement the business logic for each instruction in your program.

Each public function in the module with the #[program] attribute will be treated as a separate instruction.

Each instruction function requires a parameter of type Context and can optionally include additional function parameters representing instruction data. Anchor will automatically handle instruction data deserialization so that you can work with instruction data as Rust types.

#[program]
mod program_module_name {
    use super::*;

    pub fn instruction_one(ctx: Context<InstructionAccounts>, instruction_data: u64) -> Result<()> {
		ctx.accounts.account_name.data = instruction_data;
        Ok(())
    }
}

Instruction Context​

The Context type exposes instruction metadata and accounts to your instruction logic.

pub struct Context<'a, 'b, 'c, 'info, T> {
    /// Currently executing program id.
    pub program_id: &'a Pubkey,
    /// Deserialized accounts.
    pub accounts: &'b mut T,
    /// Remaining accounts given but not deserialized or validated.
    /// Be very careful when using this directly.
    pub remaining_accounts: &'c [UncheckedAccount<'info>],
    /// Bump seeds found during constraint validation. This is provided as a
    /// convenience so that handlers don't have to recalculate bump seeds or
    /// pass them in as arguments.
    pub bumps: BTreeMap<String, u8>,
}

Context is a generic type where T defines the list of accounts an instruction requires. When you use Context, you specify the concrete type of T as a struct that adopts the Accounts trait (e.g. Context<AddMovieReviewAccounts>). Through this context argument the instruction can then access:

  • The accounts passed into the instruction (ctx.accounts)
  • The program ID (ctx.program_id) of the executing program
  • The remaining accounts (ctx.remaining_accounts). The remaining_accounts is a vector that contains all accounts that were passed into the instruction but are not declared in the Accounts struct.
  • The bumps for any PDA accounts in the Accounts struct (ctx.bumps)

Define instruction accounts​

The Accounts trait defines a data structure of validated accounts. Structs adopting this trait define the list of accounts required for a given instruction. These accounts are then exposed through an instruction's Context so that manual account iteration and deserialization is no longer necessary.

You typically apply the Accounts trait through the derive macro (e.g. #[derive(Accounts)]). This implements an Accounts deserializer on the given struct and removes the need to deserialize each account manually.

Implementations of the Accounts trait are responsible for performing all requisite constraint checks to ensure the accounts meet the conditions required for the program to run securely. Constraints are provided for each field using the #account(..) attribute (more on that shortly).

For example, instruction_one requires a Context argument of type InstructionAccounts. The #[derive(Accounts)] macro is used to implement the InstructionAccounts struct which includes three accounts: account_name, user, and system_program.

#[program]
mod program_module_name {
    use super::*;
    pub fn instruction_one(ctx: Context<InstructionAccounts>, instruction_data: u64) -> Result<()> {
		...
        Ok(())
    }
}

#[derive(Accounts)]
pub struct InstructionAccounts {
    #[account(init, payer = user, space = 8 + 8)]
    pub account_name: Account<'info, AccountStruct>,
    #[account(mut)]
    pub user: Signer<'info>,
    pub system_program: Program<'info, System>,

}

When instruction_one is invoked, the program:

  • Checks that the accounts passed into the instruction match the account types specified in the InstructionAccounts struct
  • Checks the accounts against any additional constraints specified

If any accounts passed into instruction_one fail the account validation or security checks specified in the InstructionAccounts struct, then the instruction fails before even reaching the program logic.

Account validation​

You may have noticed in the previous example that one of the accounts in InstructionAccounts was of type Account, one was of type Signer, and one was of type Program.

Anchor provides a number of account types that can be used to represent accounts. Each type implements different account validation. We’ll go over a few of the common types you may encounter, but be sure to look through the full list of account types.

Account​

Account is a wrapper around UncheckedAccount that verifies program ownership and deserializes the underlying data into a Rust type.

// Deserializes this info
pub struct UncheckedAccount<'a> {
    pub key: &'a Pubkey,
    pub is_signer: bool,
    pub is_writable: bool,
    pub lamports: Rc<RefCell<&'a mut u64>>,
    pub data: Rc<RefCell<&'a mut [u8]>>,    // <---- deserializes account data
    pub owner: &'a Pubkey,    // <---- checks owner program
    pub executable: bool,
    pub rent_epoch: u64,
}

Recall the previous example where InstructionAccounts had a field account_name:

pub account_name: Account<'info, AccountStruct>

The Account wrapper here does the following:

  • Deserializes the account data in the format of type AccountStruct
  • Checks that the program owner of the account matches the program owner specified for the AccountStruct type.

When the account type specified in the Account wrapper is defined within the same crate using the #[account] attribute macro, the program ownership check is against the programId defined in the declare_id! macro.

The following are the checks performed:

// Checks
Account.info.owner == T::owner()
!(Account.info.owner == SystemProgram && Account.info.lamports() == 0)

Signer​

The Signer type validates that the given account signed the transaction. No other ownership or type checks are done. You should only use the Signer when the underlying account data is not required in the instruction.

For the user account in the previous example, the Signer type specifies that the user account must be a signer of the instruction.

The following check is performed for you:

// Checks
Signer.info.is_signer == true

Program​

The Program type validates that the account is a certain program.

For the system_program account in the previous example, the Program type is used to specify the program should be the system program. Anchor provides a System type which includes the programId of the system program to check against.

The following checks are performed for you:

//Checks
account_info.key == expected_program
account_info.executable == true

Add constraints with #[account(..)]​

The #[account(..)] attribute macro is used to apply constraints to accounts. We'll go over a few constraint examples in this and future lessons, but at some point be sure to look at the full list of possible constraints.

Recall again the account_name field from the InstructionAccounts example.

#[account(init, payer = user, space = 8 + 8)]
pub account_name: Account<'info, AccountStruct>,
#[account(mut)]
pub user: Signer<'info>,

Notice that the #[account(..)] attribute contains three comma-separated values:

  • init - creates the account via a CPI to the system program and initializes it (sets its account discriminator)
  • payer - specifies the payer for the account initialization to be the user account defined in the struct
  • space- specifies that the space allocated for the account should be 8 + 8 bytes. The first 8 bytes are for a discriminator that Anchor automatically adds to identify the account type. The next 8 bytes allocate space for the data stored on the account as defined in the AccountStruct type.

For user we use the #[account(..)] attribute to specify that the given account is mutable. The user account must be marked as mutable because lamports will be deducted from the account to pay for the initialization of account_name.

#[account(mut)]
pub user: Signer<'info>,

Note that the init constraint placed on account_name automatically includes a mut constraint so that both account_name and user are mutable accounts.

#[account]​

The #[account] attribute is applied to structs representing the data structure of a Solana account. It implements the following traits:

  • AccountSerialize
  • AccountDeserialize
  • AnchorSerialize
  • AnchorDeserialize
  • Clone
  • Discriminator
  • Owner

You can read more about the details of each trait. However, mostly what you need to know is that the #[account] attribute enables serialization and deserialization, and implements the discriminator and owner traits for an account.

The discriminator is an 8-byte unique identifier for an account type derived from the first 8 bytes of the SHA256 hash of the account type's name. The first 8 bytes are reserved for the account discriminator when implementing account serialization traits (which is almost always in an Anchor program).

As a result, any calls to AccountDeserialize’s try_deserialize will check this discriminator. If it doesn’t match, an invalid account was given, and the account deserialization will exit with an error.

The #[account] attribute also implements the Owner trait for a struct using the programId declared by declareId of the crate #[account] is used in. In other words, all accounts initialized using an account type defined using the #[account] attribute within the program are also owned by the program.

As an example, let's look at AccountStruct used by the account_name of InstructionAccounts

#[derive(Accounts)]
pub struct InstructionAccounts {
    #[account(init, payer = user, space = 8 + 8)]
    pub account_name: Account<'info, AccountStruct>,
    ...
}

#[account]
pub struct AccountStruct {
    data: u64
}

The #[account] attribute ensures that it can be used as an account in InstructionAccounts.

When the account_name account is initialized:

  • The first 8 bytes is set as the AccountStruct discriminator
  • The data field of the account will match AccountStruct
  • The account owner is set as the programId from declare_id

Bring it all together​

When you combine all of these Anchor types you end up with a complete program. Below is an example of a basic Anchor program with a single instruction that:

  • Initializes a new account
  • Updates the data field on the account with the instruction data passed into the instruction
// Use this import to gain access to common anchor features
use anchor_lang::prelude::*;

// Program onchain address
declare_id!("Fg6PaFpoGXkYsidMpWTK6W2BeZ7FEfcYkg476zPFsLnS");

// Instruction logic
#[program]
mod program_module_name {
    use super::*;
    pub fn instruction_one(ctx: Context<InstructionAccounts>, instruction_data: u64) -> Result<()> {
        ctx.accounts.account_name.data = instruction_data;
        Ok(())
    }
}

// Validate incoming accounts for instructions
#[derive(Accounts)]
pub struct InstructionAccounts<'info> {
    #[account(init, payer = user, space = 8 + 8)]
    pub account_name: Account<'info, AccountStruct>,
    #[account(mut)]
    pub user: Signer<'info>,
    pub system_program: Program<'info, System>,

}

// Define custom program account type
#[account]
pub struct AccountStruct {
    data: u64
}

You are now ready to build your own Solana program using the Anchor framework!

Lab​

Before we begin, install Anchor by following the steps from the Anchor docs.

For this lab we'll create a simple counter program with two instructions:

  • The first instruction will initialize an account to store our counter
  • The second instruction will increment the count stored in the counter

1. Setup​

Create a new project called anchor-counter by running anchor init:

anchor init anchor-counter

Change into the new directory, then run anchor build

cd anchor-counter
anchor build

Anchor build will also generate a keypair for your new program - the keys are saved in the target/deploy directory.

Open the file lib.rs and look at declare_id!:

declare_id!("BouTUP7a3MZLtXqMAm1NrkJSKwAjmid8abqiNjUyBJSr");

Run anchor keys sync

anchor keys sync

You'll see the Anchor updates both:

  • The key used in declare_id!() in lib.rs
  • The key in Anchor.toml

To match the key generated during anchor build:

Found incorrect program id declaration in "anchor-counter/programs/anchor-counter/src/lib.rs"
Updated to BouTUP7a3MZLtXqMAm1NrkJSKwAjmid8abqiNjUyBJSr

Found incorrect program id declaration in Anchor.toml for the program `anchor_counter`
Updated to BouTUP7a3MZLtXqMAm1NrkJSKwAjmid8abqiNjUyBJSr

All program id declarations are synced.

Finally, delete the default code in lib.rs until all that is left is the following:

use anchor_lang::prelude::*;

declare_id!("your-private-key");

#[program]
pub mod anchor_counter {
    use super::*;

}

2. Implement Counter​

First, let's use the #[account] attribute to define a new Counter account type. The Counter struct defines one count field of type u64. This means that we can expect any new accounts initialized as a Counter type to have a matching data structure. The #[account] attribute also automatically sets the discriminator for a new account and sets the owner of the account as the programId from the declare_id! macro.

#[account]
pub struct Counter {
    pub count: u64,
}

3. Implement Context type Initialize​

Next, using the #[derive(Accounts)] macro, let’s implement the Initialize type that lists and validates the accounts used by the initialize instruction. It'll need the following accounts:

  • counter - the counter account initialized in the instruction
  • user - payer for the initialization
  • system_program - the system program is required for the initialization of any new accounts
#[derive(Accounts)]
pub struct Initialize<'info> {
    #[account(init, payer = user, space = 8 + 8)]
    pub counter: Account<'info, Counter>,
    #[account(mut)]
    pub user: Signer<'info>,
    pub system_program: Program<'info, System>,
}

4. Add the initialize instruction​

Now that we have our Counter account and Initialize type , let’s implement the initialize instruction within #[program]. This instruction requires a Context of type Initialize and takes no additional instruction data. In the instruction logic, we are simply setting the counter account’s count field to 0.

pub fn initialize(ctx: Context<Initialize>) -> Result<()> {
    let counter = &mut ctx.accounts.counter;
    counter.count = 0;
    msg!("Counter Account Created");
    msg!("Current Count: { }", counter.count);
    Ok(())
}

5. Implement Context type Update​

Now, using the #[derive(Accounts)] macro again, let’s create the Update type that lists the accounts that the increment instruction requires. It'll need the following accounts:

  • counter - an existing counter account to increment
  • user - payer for the transaction fee

Again, we’ll need to specify any constraints using the #[account(..)] attribute:

#[derive(Accounts)]
pub struct Update<'info> {
    #[account(mut)]
    pub counter: Account<'info, Counter>,
    pub user: Signer<'info>,
}

6. Add increment instruction​

Lastly, within #[program], let’s implement an increment instruction to increment the count once a counter account is initialized by the first instruction. This instruction requires a Context of type Update (implemented in the next step) and takes no additional instruction data. In the instruction logic, we are simply incrementing an existing counter account’s count field by 1.

pub fn increment(ctx: Context<Update>) -> Result<()> {
    let counter = &mut ctx.accounts.counter;
    msg!("Previous counter: {}", counter.count);
    counter.count = counter.count.checked_add(1).unwrap();
    msg!("Counter incremented. Current count: {}", counter.count);
    Ok(())
}

7. Build​

All together, the complete program will look like this:

use anchor_lang::prelude::*;

declare_id!("BouTUP7a3MZLtXqMAm1NrkJSKwAjmid8abqiNjUyBJSr");

#[program]
pub mod anchor_counter {
    use super::*;

    pub fn initialize(ctx: Context<Initialize>) -> Result<()> {
        let counter = &mut ctx.accounts.counter;
        counter.count = 0;
        msg!("Counter account created. Current count: {}", counter.count);
        Ok(())
    }

    pub fn increment(ctx: Context<Update>) -> Result<()> {
        let counter = &mut ctx.accounts.counter;
        msg!("Previous counter: {}", counter.count);
        counter.count = counter.count.checked_add(1).unwrap();
        msg!("Counter incremented. Current count: {}", counter.count);
        Ok(())
    }
}

#[derive(Accounts)]
pub struct Initialize<'info> {
    #[account(init, payer = user, space = 8 + 8)]
    pub counter: Account<'info, Counter>,
    #[account(mut)]
    pub user: Signer<'info>,
    pub system_program: Program<'info, System>,
}

#[derive(Accounts)]
pub struct Update<'info> {
    #[account(mut)]
    pub counter: Account<'info, Counter>,
    pub user: Signer<'info>,
}

#[account]
pub struct Counter {
    pub count: u64,
}

Run anchor build to build the program.

8. Testing​

Anchor tests are typically Typescript integration tests that use the mocha test framework. We'll learn more about testing later, but for now navigate to anchor-counter.ts and replace the default test code with the following:

import * as anchor from "@coral-xyz/anchor";
import { Program } from "@coral-xyz/anchor";
import { expect } from "chai";
import { AnchorCounter } from "../target/types/anchor_counter";

describe("anchor-counter", () => {
  // Configure the client to use the local cluster.
  const provider = anchor.AnchorProvider.env();
  anchor.setProvider(provider);

  const program = anchor.workspace.AnchorCounter as Program<AnchorCounter>;

  const counter = anchor.web3.Keypair.generate();

  it("Is initialized!", async () => {});

  it("Incremented the count", async () => {});
});

The above code generates a new keypair for the counter account we'll be initializing and creates placeholders for a test of each instruction.

Next, create the first test for the initialize instruction:

it("Is initialized!", async () => {
  // Add your test here.
  const tx = await program.methods
    .initialize()
    .accounts({ counter: counter.publicKey })
    .signers([counter])
    .rpc();

  const account = await program.account.counter.fetch(counter.publicKey);
  expect(account.count.toNumber()).to.equal(0);
});

Next, create the second test for the increment instruction:

it("Incremented the count", async () => {
  const tx = await program.methods
    .increment()
    .accounts({ counter: counter.publicKey, user: provider.wallet.publicKey })
    .rpc();

  const account = await program.account.counter.fetch(counter.publicKey);
  expect(account.count.toNumber()).to.equal(1);
});

Lastly, run anchor test and you should see the following output:

anchor-counter
✔ Is initialized! (290ms)
✔ Incremented the count (403ms)


2 passing (696ms)

Running anchor test automatically spins up a local test validator, deploys your program, and runs your mocha tests against it. Don't worry if you're confused by the tests for now - we'll dig in more later.

Congratulations, you just built a Solana program using the Anchor framework! Feel free to reference the solution code if you need some more time with it.

Challenge​

Now it’s your turn to build something independently. Because we're starting with simple programs, yours will look almost identical to what we just created. It's useful to try and get to the point where you can write it from scratch without referencing prior code, so try not to copy and paste here.

  1. Write a new program that initializes a counter account
  2. Implement both an increment and decrement instruction
  3. Build and deploy your program like we did in the lab
  4. Test your newly deployed program and use Solana Explorer to check the program logs

As always, get creative with these challenges and take them beyond the basic instructions if you want - and have fun!

Try to do this independently if you can! But if you get stuck, feel free to reference the solution code.

Push your code to GitHub and tell us what you thought of this lesson!