Updates

October

💬 Talk

Journey of a bug at ACDT proposed adopting formal test case documentation with input from security researchers, client developers, and EIP authors. This kind of collaboration catches edge cases early and meaningfully increases test coverage.

⛽ Gas benchmarking

Pitstop was started to treat gas benchmarking config as first-class citizens of the protocol.

💻 DevEx

PokeBAL can now filter only failing tests.
Block access list progress can be tracked on forkcast.org using data from the adoption tracker.

Forkcast tracks BAL progress

🦖 FOCIL

Helping with execution tests for FOCIL, currently reviewing the specs and proposed test cases.

🎱 Block Level Access Lists

bal@v1.2.0 added test cases for no-op, 7702, 2930, and net-zero balance scenarios, followed by a patch release. Clients are passing all current tests, though they still need to map error messages to match the testing framework.

An upcoming release covers coinbase, withdrawals, precompiles, and edge cases for storage and account access.

Track progress on PokeBAL.

🔭 Experiments

Continued experimenting with EIP versioning using GitHub CI, focusing on EIP-7577 adoption with minimal overhead for EIP authors.
Started Upsilon, a minimal stateless Ethereum Virtual Machine to help prototype proposals and evaluate resource usage (compute and memory). Researching implementation strategy and language.

September

🧪 Testing Block Level Access Lists

Since the last test release, focus has been on more complex test cases.

These tests revealed that the intersection of Block Access Lists and some existing EIPs (such as 7702) is quite nuanced.

For instance, when creating a 7702 delegation, the target account doesn't need to be read from the state for the delegation to be created because the target address can be retrieved directly from the transaction. This has been clarified in the EIP and a negative test case was added to ensure that clients do not perform a state read for the target account when creating the delegation.

System operations like SELFDESTRUCT and precompile calls also needed to be addressed.

Throughout this month, several bugs within the execution specs have been fixed, and the EIP has been refined to cover learnings and all the edge cases identified during testing.

Around 30 new test cases are currently under review, and a new release BAL v1.1.0 is now available.

Geth is passing about 88% of released tests.

Track progress on PokeBAL.

🔭 Experiments

An early look into Guillotine looks promising because it's extremely fast, an important tool to have for complex prototyping.

Exploring automation around EIP-7577 semantic versioning for EIPs using GitHub Actions. The idea is to automatically track EIP changes and apply a version. WIP.