github onflow/cadence v0.5.0

latest releases: v1.0.0-preview-atree-register-inlining.21, v1.0.0-preview.21, v0.42.11-atree-register-inlining...
3 years ago

⭐ Features and Improvements

Crypto

A new built-in contract Crypto was added for performing cryptographic operations.

The contract can be imported using import Crypto.

This first iteration provides support for validating signatures, with an API that supports multiple signatures and weighted keys.

For example, to verify two signatures with equal weights for some signed data:

let keyList = Crypto.KeyList()

let publicKeyA = Crypto.PublicKey(
    publicKey:
        "db04940e18ec414664ccfd31d5d2d4ece3985acb8cb17a2025b2f1673427267968e52e2bbf3599059649d4b2cce98fdb8a3048e68abf5abe3e710129e90696ca".decodeHex(),
    signatureAlgorithm: Crypto.ECDSA_P256
)
keyList.add(
    publicKeyA,
    hashAlgorithm: Crypto.SHA3_256,
    weight: 0.5
)

let publicKeyB = Crypto.PublicKey(
    publicKey:
        "df9609ee588dd4a6f7789df8d56f03f545d4516f0c99b200d73b9a3afafc14de5d21a4fc7a2a2015719dc95c9e756cfa44f2a445151aaf42479e7120d83df956".decodeHex(),
    signatureAlgorithm: Crypto.ECDSA_P256
)
keyList.add(
    publicKeyB,
    hashAlgorithm: Crypto.SHA3_256,
    weight: 0.5
)

let signatureSet = [
    Crypto.KeyListSignature(
        keyIndex: 0,
        signature:
            "8870a8cbe6f44932ba59e0d15a706214cc4ad2538deb12c0cf718d86f32c47765462a92ce2da15d4a29eb4e2b6fa05d08c7db5d5b2a2cd8c2cb98ded73da31f6".decodeHex()
    ),
    Crypto.KeyListSignature(
        keyIndex: 1,
        signature:
            "bbdc5591c3f937a730d4f6c0a6fde61a0a6ceaa531ccb367c3559335ab9734f4f2b9da8adbe371f1f7da913b5a3fdd96a871e04f078928ca89a83d841c72fadf".decodeHex()
    )
]

// "foo", encoded as UTF-8, in hex representation
let signedData = "666f6f".decodeHex()

let isValid = keyList.isValid(
    signatureSet: signatureSet,
    signedData: signedData
)

Run-time Types

The type Type was added to represent types at run-time.

To create a type value, use the constructor function Type<T>(), which accepts the static type as a type argument.

This is similar to e.g. T.self in Swift, T::class in Kotlin, and T.class in Java.

For example, to represent the type Int at run-time:

let intType: Type = Type<Int>()

This works for both built-in and user-defined types. For example, to get the type value for a resource:

resource Collectible {}

let collectibleType = Type<@Collectible>()

The function fun isInstance(_ type: Type): Bool can be used to check if a value has a certain type:

let collectible <- create Collectible()
let collectibleType = Type<@Collectible>()
let result = collectible.isInstance(collectibleType)

For example, this allows implementing a marketplace sale resource:

pub resource SimpleSale {

    pub var objectForSale: @AnyResource?
    pub let priceForObject: UFix64
    pub let requiredCurrency: Type
    pub let paymentReceiver: Capability<&{FungibleToken.Receiver}>

    init(
        objectForSale: @AnyResource,
        priceForObject: UFix64,
        requiredCurrency: Type,
        paymentReceiver: Capability<&{FungibleToken.Receiver}>
    ) {
        self.objectForSale <- objectForSale
        self.priceForObject = priceForObject
        self.requiredCurrency = requiredCurrency
        self.paymentReceiver = paymentReceiver
    }

    destroy() {
        destroy self.objectForSale
    }

    pub fun buyObject(purchaseAmount: @FungibleToken.Vault): @AnyResource {
        pre {
            self.objectForSale != nil
            purchaseAmount.balance >= self.priceForObject
            purchaseAmount.isInstance(self.requiredCurrency)
        }

        let receiver = self.paymentReceiver.borrow()
            ?? panic("failed to borrow payment receiver capability")

        receiver.deposit(from: <-purchaseAmount)
        let objectForSale <- self.objectForSale <- nil
        return <-objectForSale
    }
}

New Parser

The existing parser was replaced by a new implementation, completely written from scratch, producing the same result as the old parser.

It is significantly faster than the old parser. For example, the following benchmark shows the difference for parsing all fungible and non-fungible token contracts and example transactions:

$ go run ./cmd/parse -bench ../../flow-{n,}ft/{contracts,transactions}/*
flow-nft/contracts/ExampleNFT.cdc:          [old]        9  111116110 ns/op
flow-nft/contracts/ExampleNFT.cdc:          [new]     2712     463483 ns/op
flow-nft/contracts/NonFungibleToken.cdc:    [old]      393    3097172 ns/op
flow-nft/contracts/NonFungibleToken.cdc:    [new]     3489     348496 ns/op
flow-nft/transactions/mint_nft.cdc:         [old]      700    1574730 ns/op
flow-nft/transactions/mint_nft.cdc:         [new]    12770      94070 ns/op
flow-nft/transactions/read_nft_data.cdc:    [old]      994    1222887 ns/op
flow-nft/transactions/read_nft_data.cdc:    [new]    15242      79295 ns/op
flow-nft/transactions/setup_account.cdc:    [old]     1281     879751 ns/op
flow-nft/transactions/setup_account.cdc:    [new]    16675      71759 ns/op
flow-nft/transactions/transfer_nft.cdc:     [old]       72   16417568 ns/op
flow-nft/transactions/transfer_nft.cdc:     [new]    10000     109380 ns/op
flow-ft/contracts/CustodialDeposit.cdc:     [old]       18   64938763 ns/op
flow-ft/contracts/CustodialDeposit.cdc:     [new]     3482     354662 ns/op
flow-ft/contracts/FlowToken.cdc:            [old]        7  177111544 ns/op
flow-ft/contracts/FlowToken.cdc:            [new]     1920     640557 ns/op
flow-ft/contracts/FungibleToken.cdc:        [old]      232    5324962 ns/op
flow-ft/contracts/FungibleToken.cdc:        [new]     2947     419529 ns/op
flow-ft/contracts/TokenForwarding.cdc:      [old]       44   25136749 ns/op
flow-ft/contracts/TokenForwarding.cdc:      [new]     7183     172917 ns/op
flow-ft/transactions/burn_tokens.cdc:       [old]       37   31475393 ns/op
flow-ft/transactions/burn_tokens.cdc:       [new]    11361     105932 ns/op
flow-ft/transactions/create_forwarder.cdc:  [old]      733    1636347 ns/op
flow-ft/transactions/create_forwarder.cdc:  [new]     8127     147520 ns/op
flow-ft/transactions/create_minter.cdc:     [old]     1306     923201 ns/op
flow-ft/transactions/create_minter.cdc:     [new]    15240      77666 ns/op
flow-ft/transactions/custodial_deposit.cdc: [old]       69   16504795 ns/op
flow-ft/transactions/custodial_deposit.cdc: [new]     7940     144228 ns/op
flow-ft/transactions/get_balance.cdc:       [old]     1094    1111272 ns/op
flow-ft/transactions/get_balance.cdc:       [new]    18741      65745 ns/op
flow-ft/transactions/get_supply.cdc:        [old]     1392     740989 ns/op
flow-ft/transactions/get_supply.cdc:        [new]    46008      26719 ns/op
flow-ft/transactions/mint_tokens.cdc:       [old]       72   17435128 ns/op
flow-ft/transactions/mint_tokens.cdc:       [new]     8841     124117 ns/op
flow-ft/transactions/setup_account.cdc:     [old]     1219    1089357 ns/op
flow-ft/transactions/setup_account.cdc:     [new]    13797      84948 ns/op
flow-ft/transactions/transfer_tokens.cdc:   [old]       74   17011751 ns/op
flow-ft/transactions/transfer_tokens.cdc:   [new]     9578     125829 ns/op

The new parser also provides better error messages and will allow us to provide even better error messages in the future.

The new parser is enabled by default – if you discover any problems or regressions, please report them!

Typed Capabilities

Capabilities now accept an optional type argument, the reference type the capability should be borrowed as.

If a type argument is provided, it will be used for borrow and check calls, so there is no need to provide a type argument for the calls anymore.

The function getCapability now also accepts an optional type argument, and returns a typed capability if one is provided.

For example, the following two functions have the same behaviour:

fun cap1(): &Something? {
  // The type annotation for `cap` is only added for demonstration purposes,
  // it can also be omitted, because it can be inferred from the value
  let cap: Capability = getAccount(0x1).getCapability(/public/something)!
  return cap.borrow<&Something>()
}

fun cap2(): &Something? {
  // The type annotation for `cap` is only added for demonstration purposes,
  // it can also be omitted, because it can be inferred from the value
  let cap: Capability<&Something> =
      getAccount(0x1).getCapability<&Something>(/public/something)!
  return cap.borrow()
}

Prefer a typed capability over a non-typed one to reduce uses / borrows to a simple .borrow(), instead of having to repeatedly mention the borrowed type.

Parameters for Scripts

Just like transactions, the main functions of scripts can now have parameters.

For example, a script that can be passed an integer and a string, and which logs both values, can be written as:

pub fun main(x: Int, y: String) {
    log(x)
    log(y)
}

Standard Library

The function fun toString(): String was added to all number types and addresses. It returns the textual representation of the type. A future version of Cadence will add this function to more types.

The function fun toBytes(): [UInt8] was added to Address. It returns the byte representation of the address.

The function fun toBigEndianBytes(): [UInt8] was added to all number types. It returns the big-endian byte representation of the number value.

🐞 Bug Fixes

  • Fixed the checking of return statements:
    A missing return value is now properly reported
  • Fixed the checking of function invocations:
    Resources are temporarily moved into the invoked function
  • Disabled caching of top-level programs (transactions, scripts)
  • Fixed the comparison of optional values
  • Fixed parsing of unpadded fixed point numbers
    in the JSON encoding of Cadence values (JSON-CDC)

💥 Breaking Changes

Field Types

Fields which have a non-storable type are now rejected:

Non-storable types are:

  • Functions (e.g. ((Int): Bool))
  • Accounts (AuthAccount / PublicAccount)
  • Transactions
  • Void

A future version will also make references non-storable. Instead of storing a reference, store a capability and borrow it to acquire a reference.

Byte Arrays

Cadence now represents all byte arrays as [UInt8] rather than [Int]. This affects the following functions:

  • String.decodeHex(): [Int] => String.decodeHex(): [UInt8]
  • AuthAccount.addPublicKey(publicKey: [Int]) => AuthAccount.addPublicKey(publicKey: [UInt8])
  • AuthAccount.setCode(code: [Int]) => AuthAccount.setCode(code: [UInt8])

However, array literals such as the following will still be inferred as [Int], meaning that they can't be used directly:

myAccount.addPublicKey([1, 2, 3])

Consider using String.decodeHex() for now until type inference has been improved.

Don't miss a new cadence release

NewReleases is sending notifications on new releases.