Questions about SEP102: Adjustment of Used Gas

[llwslc]

gas_used = gas_limit - gas_remained
if gas_used * 4 < gas_limit then /*the estimated gas consumption is more than four times of real gas consumption*/
    gas_used = gas_limit /* no gas fee will be returned*/
else if gas_used * 2 < gas_limit then /*the estimated gas consumption is more than two times of real gas consumption*/
    gas_used = (gas_used + gas_limit) / 2
else then /*the estimated gas consumption is no larger than two times of real gas consumption*/
    /* return the gas fee just as ethereum*/
endif
returned_gas_fee = (gas_limit - gas_used) * gas_price

This penalty can be bypassed with a small piece of code, which doesn't really make much sense. And hope to improve the accuracy of the estimated gas.

uint gasLimit = gasleft() + 21200;
...
while (true) {
    if ((gasleft() * 100) / gasLimit < 50) break;
}

[wangkui0508]

I can not fully understand your pseudo code. Can you elaborate on it?
Can you show us a demo whose gas is badly estimated? We need test cases.

[llwslc]

https://www.smartscan.cash/block/2989379

I think this is more typical.

[llwslc]

We had to add the following code to ensure that gas was sufficient.

const gasLimit = await contract.estimateGas.func();
const tx = await contract.func({ gasLimit: gasLimit.toNumber() * 1.2 });

[llwslc]

I think smartBCH is not able to correctly estimate the gas of this function.

// SPDX-License-Identifier: UNLICENSED

pragma solidity >=0.8.0;

contract testGas {
    function ignorePenalties() external   {
        uint gasLimit = gasleft() + 21200;
        while (true) {
            if ((gasleft() * 100) / gasLimit < 50) break;
        }
    }
}

[llwslc]

Some txs sent to your email, it is sent by protonmail. Maybe in the trash mail.

[wangkui0508]

Got the mail. We'll study it.

[llwslc]

https://www.smartscan.cash/block/3183469 Is this block normal?

[llwslc]

This low-level bug, made me execute 61 reverted transactions today, until I manually raised the gaslimit. What a bad day!

[wangkui0508]

Please make your question more specific. Why is this block abnormal?
SEP102 is not meant to make arbitrage easy and gas-efficient. We do not view it as a bug.

[llwslc]

Is it normal for too many transactions to fail because of insufficient gas? Now, let's say the gas estimate is 100, if you set it to 100, it may fail due to out of gas, if you set it to 120, it will execute successfully, and only consume 80.

[wangkui0508]

Well, if normal DApp's users have such a problem, we'd better figure out a way to optimize it. If only arbitrage transactions suffer from this, then it is not the first priority to optimize it. For arbitrage, fairness is more important than gas efficiency. If you have such a problem, other arbitragers have it too. So it is fair.

[llwslc]

I see, the goal of SmartBCH is that the normal dApp user can execute one or two simple transactions without problems. If some complex transactions go wrong, it doesn't really matter. Because all other complex contracts are also not executed correctly.

SmartBCH does not promise that complex contracts will be executed correctly, please write some simple contracts.

If everyone has a problem, it's not a problem, it's fair. That's a good point. Stay hungry, stay foolish, and everyone will feel fair.

There really shouldn't be any quality requirements for personal projects. Anyway, good job!

[wangkui0508]

Now complex contracts are executing correctly on smartBCH. They work just as the specifications, including SEP102. Sorry, we cannot take your words as the specification.
Feel free to reopen this issue when you find some real bugs. And hope your arbitrage goes smoothly in the future.

[llwslc]

Thanks, classifying one transaction as a normal dApp transaction or an arbitrage transaction to determine if there is a bug is the most interesting thing I learned about EVM.

[yougotwill]

We had to add the following code to ensure that gas was sufficient.

const gasLimit = await contract.estimateGas.func();
const tx = await contract.func({ gasLimit: gasLimit.toNumber() * 1.2 });

Thank you for this information. I was having the same problem.