Solidity events are the closest thing to a “print” or “console.log” statement in Ethereum. We will explain how they work and when to use them. We will also go into a lot of technical details often omitted in other resources.
Here is a minimal example to emit a solidity event.
contract ExampleContract {
// We will explain the significance of the indexed parameter later.
event ExampleEvent(address indexed sender, uint256 someValue);
function exampleFunction(uint256 someValue) public {
emit ExampleEvent(sender, someValue);
}
}
Perhaps the most well-known events are those emitted by ERC20 tokens when they are transferred. The sender, receiver, and amount are recorded in an event.
emit Transfer(from, to, amount);
Isn’t this redundant? We can already look through the past transactions to see the transfers, and then we could look into the transaction calldata to see the same information.
This is correct, one could delete events and have no effect on the business logic of the smart contract. However, this would not be an efficient way to look at history.
The Ethereum client does not have an API for listing transactions by “type,” for example “all transfer transactions of an ERC20.“ Here are your options if you want to query transactions:
getTransaction
getTransactionFromBlock
getTransactionFromBlock
API can only tell you what transactions occurred on a particular block, it cannot target a smart contract address multiple blocks.
getTransaction
can only inspect transactions you know the transaction hash for.
Events on the other hand can be retrieved much more easily.
Here are the Ethereum client options:
events
events.allEvents
getPastEvents
Each of these requires specifying the smart contract address that the querier wishes to examine and returns a subset (or all) of the events a smart contract emitted according to the query parameters specified.
Here is the key insight for why you would use Events to track transactions rather than transactions themselves: Ethereum does not provide a mechanism to get all transactions for a smart contract, but it does provide a mechanism for getting all events from a smart contract.
Why is this? Making events quickly retrievable requires additional storage overhead. If Ethereum did this for every transaction, this would make the chain considerably larger. With events, solidity programmers can be selective about what kind of information is worth paying the additional storage overhead for, to enable quick off-chain retrieval.
Here is an example of using the API described above. In this code, the client subscribes to events from a smart contract. These examples are all in Javascript.
This code triggers a callback every time an ERC20 token emits a transfer Event.
const { ethers } = require("ethers");
// const provider = your provider
const abi = [
"event Transfer(address indexed from, address indexed to, uint256 value)"
];
const tokenAddress = "0x...";
const contract = new ethers.Contract(tokenAddress, abi, provider);
contract.on("Transfer", (from, to, value, event) => {
console.log(`Transfer event detected: from=${from}, to=${to}, value=${value}`);
});
If we want to look at events retroactively, we can use the following code. In this example, we look to the past for Approval transactions in an ERC20 token.
const ethers = require('ethers');
const tokenAddress = '0x...';
const filterAddress = '0x...';
const tokenAbi = [
{
"anonymous": false,
"inputs": [
{
"indexed": true,
"name": "from",
"type": "address"
},
{
"indexed": true,
"name": "to",
"type": "address"
},
{
"indexed": false,
"name": "value",
"type": "uint256"
}
],
"name": "Transfer",
"type": "event"
}
];
const tokenContract = new ethers.Contract(tokenAddress, tokenAbi, provider);
// this line filters for Approvals for a particular address.
const filter = tokenContract.filters.Approval(filterAddress, null, null);
tokenContract.queryFilter(filter).then((events) => {
console.log(events);
});
If you wanted to look for a trade between two particular known addresses (if such a transaction exists), the ethers.js. code would be as follows:
tokenContract.filters.Transfer(address1, address2, null);
Here is a similar example in web3.js instead of ethers.js. Note that the fromBlock
and toBlock
query parameters are added (to signify we only care about events between these blocks), and we will demonstrate the ability to listen to multiple addresses being the sender. The addresses are combined with the "OR" condition.
const Web3 = require('web3');
const web3 = new Web3('https://rpc-endpoint');
const contractAddress = '0x...'; // The address of the ERC20 contract
const contractAbi = [
{
"anonymous": false,
"inputs": [
{
"indexed": true,
"name": "from",
"type": "address"
},
{
"indexed": true,
"name": "to",
"type": "address"
},
{
"indexed": false,
"name": "value",
"type": "uint256"
}
],
"name": "Transfer",
"type": "event"
}
];
const contract = new web3.eth.Contract(contractAbi, contractAddress);
const senderAddressesToWatch = ['0x...', '0x...', '0x...']; // The addresses to watch for transfers from
const filter = {
fromBlock: 0,
toBlock: 'latest',
topics: [
web3.utils.sha3('Transfer(address,address,uint256)'),
null,
senderAddressesToWatch,
]
};
contract.getPastEvents('Transfer', {
filter: filter,
fromBlock: 0,
toBlock: 'latest',
}, (error, events) => {
if (!error) {
console.log(events);
}
});
The example above works because the Approve (and Transfer) event in ERC20 sets the sender to be indexed. Here is the ERC20 Approval event declaration in Solidity.
event Approval(address indexed owner, address indexed spender, uint256 value);
If the “owner” argument was not indexed, the javascript code earlier would silently fail. The implication here is that you cannot filter for ERC20 events that have a specific value
(token amount) for the transfer, because that is not indexed. You must pull in all the events and filter them javascript side; it cannot be done in the Ethereum client.
An indexed argument for an event declaration is called a topic.
The generally accepted best practice for events is to log them whenever a potentially consequential state change happens. Some examples include:
Changing the owner of the contract
Moving ether
Conducting a trade
Not every state change requires an event. The question the Solidity developers should ask themselves is “would someone have an interest in retrieving or discovering this transaction quickly?”
This will requires some subjective judgement. Remember, an unindexed parameter cannot be searched for directly, but still might be useful data when accompanied by an indexed parameter. A good way to get an intuition for this is to look at how established codebases design their events
As a general rule of thumb, cryptocurrency amounts should not be indexed, and an address should be, but this rule should not be applied blindly. Indexes only allow you to quickly get exact values, not a range of values.
An example of this would be adding an event when tokens are minted because underlying libraries already emit this event.
Events are state changing; they alter the state of the blockchain by storing the log. Therefore, they cannot be used in view (or pure) functions.
Events are not as useful for debugging the way console.log and print are in other languages; because events themselves are state-changing, they are not emitted if a transaction reverts.
For unindexed arguments, there is no intrinsic limit to the number of arguments, though of course there are contract size and gas limits that apply. The following nonsensical example is valid solidity:
contract ExampleContract {
event Numbers(uint256, uint256, uint256, uint256, uint256, uint256, uint256, uint256);
}
Similarly, there is no intrinsic limit to the length of strings or arrays stored in a log.
However, there cannot be more than three indexed arguments (topics) in an event. An anonymous event can have 4 indexed arguments (we will cover this distinction later).
An argument with zero events is also valid.
The following events behave identically
event NewOwner(address newOwner);
event NewOwner(address);
In general, including the variable name would be ideal because the semantics behind the following example are very ambiguous (this is not how you should declare events!)
event Trade(address,address,address,uint256,uint256);
We can guess that the addresses correspond to the sender, and the token addresses, while the uint256es correspond to the amounts, but this is hard to decipher.
It is conventional to capitalize the name of an event, but the compiler does not require it. This would be a much better declaration:
event Trade(address trader, address token1, address token2, uint256 token1Amount, uint256 token2Amount);
When an event is declared in a parent contract, it can be emitted by the child contract. Events are internal and cannot be modified to be private or public.
Here is an example:
contract ParentContract {
event NewNumber(uint256 number);
function doSomething(uint256 number) public {
emit NewNumber(number);
}
}
contract ChildContract is ParentContract {
function doSomethingElse(uint256 number) public {
emit NewNumber(number); // valid
}
}
Similarly, events can be declared in an interface and used in the child, as in the following example.
interface IExampleInterface {
event Deposit(address indexed sender, uint256 amount);
}
contract ExampleContract is IExampleInterface {
function deposit() external payable {
emit Deposit(msg.sender, msg.value); // also valid
}
}
The EVM (Ethereum Virtual Machine) identifies events with the keccak256 of their signature.
For solidity versions 0.8.15 or higher, you can also retrieve the selector using the .selector member.
pragma solidity ^0.8.15;
contract ExampleContract {
event SomeEvent(uint256 blocknum, uint256 indexed timestamp);
function selector() external pure returns (bool) {
// true
return SomeEvent.selector == keccak256("SomeEvent(uint256,uint256)");
}
}
The event selector is actually a topic itself (we will discuss this further in a later section).
Marking variables as indexed or not does not change the selector.
Events can be marked as anonymous, in which case they will not have a selector. This means that client-side code cannot specifically isolate them as a subset like our earlier examples. The only way for client-side code to see an anonymous event is to get all events for the smart contract.
pragma solidity ^0.8.15;
contract ExampleContract {
event SomeEvent(uint256 blocknum, uint256 timestamp) anonymous;
function selector() public pure returns (bool) {
// ERROR: does not compile, anonymous events don't have selectors
return SomeEvent.selector == keccak256("SomeEvent(uint256,uint256)");
}
}
Because the event signature is used as one of the indexes, an anonymous function can have four indexed topics, since the function signature is “freed up” as one of the topics.
An anonymous event can have up to four indexed topics. A non-anonymous event can have up to three.
contract ExampleContract {
// valid
event SomeEvent(uint256 indexed, uint256 indexed, address indexed, address indexed) anonymous;
}
Anonymous events are rarely used in practice.
This section describes events at the assembly level of the EVM. This section can be skipped for programmers new to blockchain development.
To retrieve every transaction that has happened with a smart contract, the Ethereum client would have to scan every block, which would be an extremely heavy I/O operation; but Ethereum uses an important optimization.
Events are stored in a Bloom Filter data structure for each block. A Bloom Filter is a probabilistic set that efficiently answers if a member is in the set or not. Instead of scanning the entire block, the client can ask the bloom filter if an event was emitted in the block. This allows the client to scan the blockchain much faster to find events.
Bloom Filters are probabilistic: they sometimes incorrectly return that an item is a member of the set even if it isn’t. The more members that are stored in a Bloom Filter, the higher the chance of error, and the larger the bloom filter must be (storage wise) to compensate for this. Because of this, Ethereum doesn’t store all the transactions in a Bloom Filter. There are far fewer events than there are transactions. This keeps the storage size on the blockchain manageable.
When the client gets a positive membership response from a bloom filter, it must scan the block to verify the event took place. However, this will only happens for a tiny subset of blocks, so on average the Ethereum client saves a lot of computation by checking the bloom filter for event presence first.
In the Yul intermediate representation, the distinction between indexed arguments (topics) and unindexed arguments becomes clear.
The following yul functions are available for emitting events (and their EVM opcode bears the same name). The table is copied from the yul documentation with some simplification.
op code |
Usage |
---|---|
log0(p, s) |
log without topics and data mem[p…(p+s)) |
log1(p, s, t1) |
log with topic t1 and data mem[p…(p+s)) |
log2(p, s, t1, t2) |
log with topics t1, t2 and data mem[p…(p+s)) |
log3(p, s, t1, t2, t3) |
log with topics t1, t2, t3 and data mem[p…(p+s)) |
log4(p, s, t1, t2, t3, t4) |
log with topics t1, t2, t3, t4 and data mem[p…(p+s)) |
A log can have up to 4 topics, but a non-anonymous solidity event can have up to 3 indexed arguments. That is because the first topic is used to store the event signature. There is no opcode or Yul function for emitting more than four topics.
The unindexed parameters are simply abi encoded in the memory region [p…(p+s)) and emitted as one long byte sequence.
Recall earlier that there was no limit in principle for how many unindexed arguments an event in Solidity can have. The underlying reason is that there is no explicit limit on how long the memory region pointed to in the first two parameters of the log op code takes. There is of course, limits provided by the contract size and memory expansion gas costs.
Events are substantially cheaper than writing to storage variables. Events are not intended to be accessible by smart contracts, so the relative lack of overhead justifies a lower gas cost.
The formula for how much gas an event costs is as follows (source):
375 + 375 * num_topics + 8 * data_size + mem_expansion cost
Each event costs at least 375 gas. An additional 375 is paid for each indexed parameter. A non-anonymous event has the event selector as an indexed parameter, so that cost is included most of the time. Then we pay 8 times the number of 32 byte words written to the chain. Because this region is stored in memory before being emitted, the memory expansion cost must be accounted for also.
In general, the intuition that the more you log the more you pay in gas, is accurate.
Events are for clients to quickly retrieve transactions that may be of interest. Although they don’t alter smart contract functionality, they allow the programmer to specify which transactions should be quickly retrievable. This makes it easier for dapps to quickly summarize important information.
Events are relatively cheap gas-wise compared to storage, but the most important factor in their cost is the number of indexed parameters, assuming the coder does not use an inordinate amount of memory.
Like what you see here? See our advanced Solidity Bootcamp to learn more.
The lead image for this article was generated by HackerNoon's AI Image Generator via the prompt "Events in Solidity"