enitrat / cairo1-vaults

What was implemented

• Implementation of the ERC4626 standard in Cairo.
• Implementation of an ERC4626-based Vault in Cairo. This Vault is a smart contract that allows users to deposit an underlying token and receive vault shares. The Vault will invest the funds into Strategies(for now, 1 Vault = 1 Strategy). These strategies share the same interface allowing to invest and withdraw funds. They can be anything from lending protocols to yield farming strategies. A user can withdraw from the vault at any time. If the vault doesn't enough funds, it will withdraw from the strategies to fulfill the withdrawal request.
• Implementation of a Mock Strategy to test the vaults. This Mock strategy is a simple Strategy that has minting rights over a mock token to simulate profits/loss.

What would've been nice

• [] Implementing a Vault factory contract so that we can deploy vaults for multiple strategies easily
• [] Supporting multiple strategies in a single vault
• [] Implementing a real yield-bearing lending strategy instead of a mock one
• [] Improve the Vault by adding more complex features such as a fee system, a timelock, a floating token target, emitting more events, etc.

Problems encountered

• There is currently no efficient and scalable way of writing composable contracts in Cairo. A Vault follows the ERC4626 standard, which itself follows the ERC20 standard. Each layer of composability requires re-defining all the functions of the underlying contract. This is not only tedious, but also error-prone. It took me a while to figure out that since I re-defined the behavior of the total_assets() function in the Vault contract, I had to recursively change the implementation of all ERC4626 functions depending on total_assets(). This is very likely to introduce new errors, and demonstrates the current limitations of the contract syntax.
• While the cairo-test runner is very useful, it was at first not very useful as any test failing would only return the ENTRYPOINT_FAILED error. This was fixed and now returns all error messages in the execution stack.
• The cairo-test runner sometimes fails with a Error: Failed setting up runner. message. The reason behind this is unclear, as it seems to be non-deterministic and happens randomly in some cases, but all the time for some tests. As a consequence, a lot of test cases that previously passed had to be commented after refactoring the code.
• The StarknetJS + devnet was not satisfying either, being too slow for it to be valuable for fast iterations.