A set of utilities for cancelling bids in the ENS .eth Registrar that
were never revealed.
The code is released as-is, without warranty of any kind, including fitness for a purpose.
You're about to see some UGLY, like '0x' + event['topics'][2][-40:]
instead of web3.toChecksumAddress(...).
Seriously. This might not be something you wish to delve into.
The utilities allow to scan for placed sealed bids, and to determine which of those haven't been revealed yet. Once their expiration timestamp is reached, a call can be made to cancel such bids.
This allows implementing a bot that cleans up the blockchain and, if efficient, also rewards its operator.
An IPC connection to a local node is required.
The on-chain component is written in Solidity, with the primary consideration being low gas price for calls.
You can see it at work on a block explorer. While there, take a look at a competitor, too.
To calculate the success rate, TODO
export VIRTUAL_ENV=.virtualenv/cancelot
mkdir -p $VIRTUAL_ENV
virtualenv $VIRTUAL_ENV
source $VIRTUAL_ENV/bin/activate
pip install -r requirements.txt
pip install -e . # TODO: make a package on PyPI
pip install ipythonAwkward.
The code has been used with CPython 3.5 for a few weeks until its first release. There are no tests.
Sections that are sub-optimal are marked with TODO. Sections
that are down-right horrible are marked with FIXME.
Many functions have side-effects, such as debug messages or warnings.
cancelot in general is provided as a Python package, with a few
example scripts that use it.
To construct a state for the first time, run:
python -u oneshot.py 2>&1 | tee -i -a logs/`date +%s`.logThe result will be a lot of pickle files. Most are saved as "backups"
while the script is running. Save the newest one in the pickles
directory, and remove the rest.
For subsequent runs, use:
./run-oneshot.shThis will reuse the previous pickle.
Several "bidding strategy" helpers made it into the repository. There is little sense in removing them now, since they're not complex anyway, and it would require some major rewriting of git history, most likely complete flattening, which I find unacceptable.
Anyway, they are available in ipybench.py. The file is to be run as:
ipython -i ipybench.pyBy necessity, several lines from IPython history that constitute a minimal bot are not provided.
This is to guarantee that at least some programming must be done by an operator wishing to use this, therefore absolving me personally from being morally responsible for the horrible misfortune they're inevitably bound to run into.
Revealing the code completely would have also put me at a total disadvantage, comparing to competing closed-source implementations, whereas now the handicap is merely severe.
The provided "strategy" is, at best, break-even. It gives most of the profit to the miners. Arguably, this is what the race-to-cancel devolves to in the presence of multiple parties trying to do the cancelling.
Somewhat counter-intuitively, gas price is not the main factor - at least, not at all times. But it is one that's always present.
Other factors include node latency and level of integration into the network, as well as specific miners' algorithms for selecting which transactions get included, and in what order.
General health of the network, such as the size of the txpool, can also be said to play a role, although the effects are included in the factors mentioned above. This works more as an indicator.
Everything in this repository is licensed under AGPLv3. See LICENSE.txt.
On block 3648565:
https://etherscan.io/tx/0x6e032eed213f1a641af8cdb7fd3770984010a38c04f6c06e393b03dfca30a80d
Bid from 0x3c12c57a05780b6e97360392ce18f1ad92fbe0a7 with seal 0x844e77749af1a22536ebbe7fed28588cc4e82302096105f0309db00dd8c79256 added (block 3665634). '0x3c12c57a05780b6e97360392ce18f1ad92fbe0a7', '0x00033095b0df8983c66c84b7ff557a5b9b4705a9e22167ae748351d6357ae98b'
Bid from 0x3e1f4f4de69e7e2cec0f45153a542d6108ef81bb with seal 0x486e1b9e1e85a60199f98c945ae548c42c51b472b8842181c1d1414a01a4f97c cancd (block 3754090).
Possibly a multisig contract doing the bidding?.. Or an index collision?..
WARNING! Key not found in store, skipping bid removal! Tried:
('bytes32 not in store', '0x3e35de8f9a0f71c7891245f50a46be4e863244a4', '0x772f91f480a2292645cedee795ffd0f03793e580ba481c16ad23c7b7d0b7f1d6')
('bytes32 not in store', '0x3e35de8f9a0f71c7891245f50a46be4e863244a4', '0x0fb37976806ab1a28e0b52bc3b81a240298f3859b560a5b74c08a9fadd21b818')
WARNING! Key not found in store, skipping bid removal! Tried:
('bytes32 not in store', '0x3e35de8f9a0f71c7891245f50a46be4e863244a4', '0x9d2c5cb0cbd9397dbe1b448b1c286f93ee7f51b13c9668bffdd09a22fa3033ba')
('bytes32 not in store', '0x3e35de8f9a0f71c7891245f50a46be4e863244a4', '0x0fb37976806ab1a28e0b52bc3b81a240298f3859b560a5b74c08a9fadd21b818')