This is copy from https://polkadot.polkassembly.io/post/469
Metis will be the Wasm contract standard library developed by Patract Labs. Patract Labs will work with community forces to formulate various Wasm contract standards, develop corresponding implementations of ink! and Ask! versions, and conduct audits by professional institutions.
We believe that in the Wasm contract ecosystem, a development library similar to Openzeppelin-contracts is very necessary. This is Metis. For the sake of simplicity and readability, the patterns and meta-language provided by this library should be as consistent with Openzeppelin-contracts as possible, so as to reduce the burden on developers and effectively absorb the experience accumulated in the Solidity ecosystem.
But you need to realize that the Wasm contract development based on ink! and Ask! is very different from the EVM contract development based on Solidity. Wasm's mature system support allows developers to use a variety of different languages and tools to develop smart contracts. Based on these mature or rapid iteration platforms, developers can enjoy a large number of underlying facilities support and development experience accumulation, but different languages and tools often have different considerations and trade-offs. This means that developers will use completely different codes to express the same mechanism and design, which will bring huge communication costs and mental burdens to the community.
Faced with such problems, we need to summarize and accumulate the implementation model in the development of smart contracts. This idea was introduced in the book "Implementation Patterns" by Kent Beck. The purpose of summarizing the implementation mode is to clearly and accurately express the developer's intentions and ideas through these clear patterns, so that "code that others can understand" can be implemented. In contract development, this idea is very important.
From Openzeppelin-contracts, we can see that it summarizes several implementation models in the development of smart contracts, such as the "contract expansion model" based on the Solidity inheritance grammar. These implementation models are valuable experience accumulated in the Solidity ecology. Of course, different languages have different ways of practicing these modes. Metis will implement these modes and provide corresponding support on different platforms. For example, in Rust, it is obvious that you cannot directly use inheritance to implement contract expansion. For this, Metis will provide code implementation examples and provide a series of help libraries to reduce the developer's mental cost.
Metis will be more than just a Wasm contract standard library. We hope that through the practice of Metis, we can fully inherit and absorb previous contract development experience while exploring the best practices of Wasm contract development.
At present, contract-based developers mainly face the following problems:
The above problems severely limit the current ink!-based contract development ecology. Metis will solve these problems while avoiding the existing problems in Solidity.
Metis will implement a series of common components, similar to the Openzeppelin-contracts development library. These components will be thoroughly tested and code audited. These components will be as consistent as possible with Openzeppelin-contracts, which can reduce the burden on developers and effectively absorb the experience accumulated in the Solidity ecosystem.
Metis will include the following components:
In the previous version of Metis, we will first implement Openzeppelin-contracts-like components for developers to use. These components include:
Most of Metis development libraries are composed of contract components. In Solidity, the introduction of contract components can be implemented based on inheritance. Generally, the contract components will include the following parts:
For example, common Ownable contracts:
abstract contract Ownable is Context {
// Component Storage
address private _owner;
// Component Event
event OwnershipTransferred(address indexed previousOwner, address indexed newOwner);
// Component internal interface
modifier onlyOwner() {
...
}
// Component Event
function renounceOwnership() public virtual onlyOwner {
...
}
...
}
If a contract needs to be Ownable, just inherit the contract:
contract Escrow is Ownable {
...
}
Similarly, other components can also have Ownable through inheritance:
contract Pausable is Ownable {
...
function unpause() public onlyOwner whenPaused {
...
}
}
The intention of inheritance here is actually composition rather than an is-a relationship. Although excessive use of inheritance in Solidity will cause many problems, inheritance for specific intentions is still an important way to achieve it.
In contract development based on ink!, through metis, we can also achieve the same motivation based on generics and traits:
...
#[metis::contract] // metis contract macro, will use ink_lang::contract auto
mod flipper {
...
#[ink(storage)]
#[import(ownable)] // flipper import the ownable
pub struct Flipper {
ownable: ownable::Data<Flipper>, // data by ownable
value: bool,
}
#[ink(event)]
#[metis(ownable)] // TODO: event in ink! will refactor
pub struct OwnershipTransferred {
...
}
impl Flipper {
...
#[ink(message)]
pub fn flip(&mut self) {
// check owner
ownable::Impl::ensure_caller_is_owner(self);
self.value = !self.value;
}
...
#[ink(message)]
pub fn renounce_ownership(&mut self) {
ownable::Impl::renounce_ownership(self) // owner message
}
...
}
}
Metis assists developers to achieve the same function through a series of helper macros. In order to improve the auditability of the contract, here we hope that users clearly implement storage, event and message declarations.
At the same time, metis has improved the implementation of components:
...
// Storage
#[metis::component::storage]
pub struct Data<E>
where
E: Env,
{
owner: Lazy<Option<E::AccountId>>,
}
// Event trait
pub trait EventEmit<E: Env>: EnvAccess<E> {
...
}
// Impl trait
pub trait Impl<E: Env>: Storage<E> + EventEmit<E> {
// logics
fn init(&mut self) {
self.get_mut().set_ownership(&Some(Self::caller()));
}
fn renounce_ownership(&mut self) {
self.ensure_caller_is_owner();
self.emit_event_ownership_transferred(
self.get().get_ownership().clone(),
None);
self.get_mut().set_ownership(&None);
}
...
/// Panic if `owner` is not an owner
fn ensure_owner(&self, owner: &E::AccountId) {
assert!(&self.get().get_ownership().clone().unwrap() == owner);
}
...
}
Such components can extend their functions by inheriting other components, such as an ERC20 component with the function of destroying tokens:
...
pub trait Impl<E>: erc20::Impl<E>
where
E: Env,
{
fn _burn(&mut self, account: &E::AccountId, amount: E::Balance) -> Result<()> {
...
}
fn burn(&mut self, amount: E::Balance) -> Result<()> {
self._burn(&Self::caller(), amount)
}
fn burn_from(&mut self, account: &E::AccountId, amount: E::Balance) -> Result<()> {
...
}
}
Based on metis, we can implement various contract combination modes implemented by Solidity through inheritance under limited intent, and at the same time, with the help of rust's zero-cost abstraction, these abstractions will not bring additional performance consumption.
Metis for ink! is divided into several milestones:
Considering that the current ink! and contract-pallet are still in iteration, some metis features will be implemented based on subsequent improvements, including:
With the richness and completeness of the ink! contract community, metis will further implement more public components and libraries to assist developers in developing large-scale contract projects. Therefore, we may arrange [MR] milestones. It will be developed based on the iterative schedule of ink!
Cost of M1 (15 developers * weeks):
Complete the macros and tests related to the metis component, and all the tests can pass. Complete the detailed metis component design document. Complete the implementation and testing of the following contract components, all tests can pass:
Complete the documentation and examples of the above contract components.
Metis is a development library for Wasm contract ecosystem, similar to Openzeppelin-contracts in Solidity. It will provide a series of reliable public components to implement contracts and summarize the implementation model in the development of smart contracts.
Metis will implement a series of common components, including basic Access and Security components, as well as Token and Governance components. It will also include ink! components, summarizing abstract components from mature contract projects, and the expansion of ink! contracts.
Metis will be developed in several milestones, with M1 focusing on implementing basic component macros and components, improving component testing support, and completing the implementation and testing of Access, Security, and Token related components.
The cost of M1 is estimated to be $31,500, and it will be verified by completing the macros and tests related to the metis component and passing all tests, completing the detailed metis component design document, and completing the implementation and testing of the above contract components.