[ISSUE-121] docs: research gas and cost efficiency for event emission

[0xjoaovpsantos]

docs: research gas and cost efficiency for event emission

closes #121

How I calculated the gas usage?

• Example:

uint256 startGas = gasleft();

// code

gasUsed = startGas - gasleft();

Prepare a spreadsheet with the gas usage based on the current and proposed events emitted and set for a comparison.

• In the document I separeted a comparison about the options to emit events :)

Estimate the gas when calling createSwap and record their gas usage.

• When we call the "createSwap" event with the current parameters (id, owner and expiry) the gas usage is 1932.

event SwapCreated(
    uint256 indexed id,
    address indexed owner,
    uint256 indexed expiry
);

emit SwapCreated(swapId, msg.sender, swap.expiry);

• But we can optimize with assembly, so the gas usage can be: 1919

//keccak-256(SwapCreated(uint256,address,uint256))
bytes32 private constant EVENT_SWAP_CREATED_SIGNATURE =
    0xecac413765eac3982e66e569ad3a200e6b3f188ac4d7d44cc5da667779ca8d05;

uint256 swapExpiry = swap.expiry;
assembly {
    log4(0x00, 0x00, EVENT_SWAP_CREATED_SIGNATURE, swapId, caller(), swapExpiry)
}

Take measures of the gas usage.

• I change the emit of events to use assembly to reduce the gas usage.

Add the allowed field in the events. Take Measure.

• Gas usage with Solidity: 2702

event SwapCreated(
    uint256 indexed id,
    address indexed owner,
    uint256 indexed expiry,
    address allowed
);

emit SwapCreated(swapId, msg.sender, swap.expiry, swap.allowed);

• But we can optimize with assembly, so the gas usage can be: 2502

//keccak-256(SwapCreated(uint256,address,uint256,address))
bytes32 private constant EVENT_SWAP_CREATED_SIGNATURE =
    0x43a58bfac3282e5ce3bfd714c5bc0ddff8d6f2cd049db0b02a25d7cdd5026efb;

uint256 swapExpiry = swap.expiry;
address allowed = swap.allowed;
assembly {
    mstore(0x00, allowed)
    log4(0x00, 0x20, EVENT_SWAP_CREATED_SIGNATURE, swapId, caller(), swapExpiry)
}

Add the bidding and asking field in the events. Take Measure.

• Gas usage with Solidity: 7646

event SwapCreated(
    uint256 indexed id,
    address indexed owner,
    uint256 indexed expiry,
    address allowed,
    ISwap.Asset[] biding,
    ISwap.Asset[] asking
);

emit SwapCreated(swapId, msg.sender, swap.expiry, swap.allowed, swap.biding, swap.asking);

Add the price of ETH at $2.200 and calculate the gas cost in dollars.

• I don't know how I can make this setting the ETH value. (help me please :)

Conclusion

• After the study about gas usage in the swaplace, I came to the conclusion that we can add the "allowed" field in the "SwapCreated" event, but if we add the "biding" and "asking" fields, we will increase a lot the gas usage in the emit of events, so I this PR I added the "allowed" field in the swap created event and refactored all events to use assembly to reduce the gas usage.

• And about the "biding" and "asking" fields, I think that is have a lot of gas usage and I'm studying about how we can emit the event with assembly there paramters, I tried but I couldn't, I'm trying yet :)

• If possible, I need your review about this PR :), I'm ready to fix/improve something and work togheter!!!

[0xneves]

Let's go @0xjoaovpsantos 🔥

Hitting some scores: 1 - Using assembly to optimize will turn legibility in the opposite direction we expect. 2 - Allowed should be indexed instead of expiry as it is more important for tracking enrolled users in the swap. 2.1 - Event emission parameters should follow the same order as the Swap struct. 3 - The comparison table was lacking from the response so I had to test before and after changes. 3.1 - Adding allowed in the createSwap event will add up to ~147 gas. 4 - Logging the bidding and asking assets will slightly increase gas consumption. 4.1 - Logging structs in assembly is a pain in the ass, I went after the resolution, and it is just not worth it.

I applied this PR resolution in my fork while comparing the gas differences. Please feel free to take a look, and commit in your branch, and we can move forward with this PR conclusion.

Nice job 😄

[0xjoaovpsantos]

Thanky you @0xneves !!!!

• I updated this PR with your recommendation in your branch refactor-logs: https://github.com/0xneves/swaplace-contracts/tree/refactor-logs

• Sorry I forgot the compariosion table in the readme of PR :/, I'll improve in the next issues :)