Help is always welcome and there are plenty of options to contribute to the Ethereum Execution Layer Specifications (EELS).
In particular, we appreciate support in the following areas:
- Reporting issues
- Fixing and responding to issues, especially those tagged as E-easy which are meant as introductory issues for external contributors.
- Improving the documentation.
For details about EELS usage and building, please refer the README
This specification aims to be:
- Correct - Describe the intended behavior of the Ethereum blockchain, and any deviation from that is a bug.
- Complete - Capture the entirety of consensus critical parts of Ethereum.
- Accessible - Prioritize readability, clarity, and plain language over performance and brevity.
- Attempt to use descriptive English words (or very common abbreviations) in documentation and identifiers.
- Avoid using EIP numbers in identifiers.
- If necessary, there is a custom dictionary
whitelist.txt
.
Running the tests necessary to merge into the repository requires:
- Python 3.9.x, and
- PyPy 7.3.x.
geth
installed and present in$PATH
execution-specs
depends on a submodule that contains common tests that are run across all clients, so we need to clone the repo with the --recursive flag. Example:
$ git clone --recursive https://github.com/ethereum/execution-specs.git
Or, if you've already cloned the repository, you can fetch the submodules with:
$ git submodule update --init --recursive
The tests can be run with:
$ tox
The development tools can also be run outside of tox
, and can automatically reformat the code:
$ pip install -e ".[doc,lint,test]" # Installs ethereum, and development tools.
$ isort src # Organizes imports.
$ black src # Formats code.
$ flake8 # Reports style/spelling/documentation errors.
$ mypy src # Verifies type annotations.
$ pytest -n 4 # Runs tests parallelly.
$ pytest -m "not slow" # Runs tests which execute quickly.
It is recommended to use a virtual environment to keep your system Python installation clean.
A trace of the EVM execution for any test case can be obtained by providing the --evm-trace
argument to pytest.
Note: Make sure to run the EVM trace on a small number of tests at a time. The log might otherwise get very big.
Below is an example.
pytest tests/frontier/test_state_transition.py -k 'test_general_state_tests_new' --evm-trace
The EELS repository has various CLI utilities that can help in the development process.
This tool can be used to create the base code for a new fork by using the existing code from a given fork.
The command takes 4 arguments, 2 of which are optional
- from_fork: The fork name from which the code is to be duplicated. Eg. - "Tangerine Whistle"
- to_fork: The fork name of the new fork Eg - "Spurious Dragon"
- from_test (Optional): Name of the from fork within the test fixtures in case it is different from fork name. Eg. - "EIP150"
- to_test (Optional): Name of the to fork within the test fixtures in case it is different from fork name Eg - "EIP158"
As an example, if one wants to create baseline code for the Spurious Dragon
fork from the Tangerine Whistle
one
ethereum-spec-new-fork --from_fork="Tangerine Whistle" --to_fork="Spurious Dragon" --from_test=EIP150 --to_test=EIP158
The following will have to however, be updated manually
- The fork number and
MAINNET_FORK_BLOCK
in__init__.py
. If you are proposing a new EIP, please setMAINNET_FORK_BLOCK
toNone
. - Any absolute package imports from other forks eg. in
trie.py
- Package names under
setup.cfg
- Add the new fork to the
monkey_patch()
function insrc/ethereum_optimized/__init__.py
- Adjust the underline in
fork/__init__.py
The sync tool allows one to use an RPC provider to fetch and validate blocks against EELS. The state can also be stored in a local DB after validation. Since syncing directly with the specs can be very slow, one can also leverage the optimized module. This contains alternative implementations of routines in EELS that have been optimized for speed rather than clarity/readability.
The tool can be called using the ethereum-spec-sync
command which takes the following arguments
- rpc-url: Endpoint providing the Ethereum RPC API. Defaults to
http://localhost:8545/
- unoptimized: Don't use the optimized state/ethash (this can be extremely slow)
- persist: Store the state in a db in this file
- geth: Use geth specific RPC endpoints while fetching blocks
- reset: Delete the db and start from scratch
- gas-per-commit: Commit to db each time this much gas is consumed. Defaults to 1_000_000_000
- initial-state: Start from the state in this db, rather than genesis
- stop-at: After syncing this block, exit successfully
- The following options are not supported WITH
--unoptimized
->--persist
,--initial-state
,--reset
- The following options are not supported WITHOUT
--persist
->--initial_state
,--reset
This tool can be used to apply the unstaged changes in SOURCE_FORK
to each of the TARGET_FORKS
. If some
of the change didn't apply, '.rej' files listing the unapplied changes will be left in the TARGET_FORK
.
The tool takes the following command line arguments
- The fork name where the changes have been made. Eg:-
frontier
(only a single fork name) - The fork names where the changes have to be applied. Eg:-
homestead
(multiple values can be provided separated by space) - optimized: Patch the optimized code instead
- tests: Patch the tests instead
As an example, if one wants to apply changes made in Frontier
fork to Homestead
and Tangerine Whistle
python src/ethereum_spec_tools/patch_tool.py frontier homestead tangerine_whistle
This tool checks for style and formatting issues specific to EELS and emits diagnostics when issues are found
The tool currently performs the following checks
- The order of the identifiers between each hardfork is consistent.
- Import statements follow the relevant import rules in modules.
The command to run the tool is ethereum-spec-lint