exit queue introspection

[InoMurko]

The test adds two exit queues:
16:44:55.122 [info] ETH exit queue missing. Adding...
16:44:57.149 [info] ERC20 exit queue missing. Adding...
after which I immidiately call: getNextExit(uint256,address)

data =
ABI.encode("getNextExit(uint256,address)", [Itest.PlasmaFramework.vault_id(Currency.ether()), Currency.ether()])

{:ok, result} = Ethereumex.HttpClient.eth_call(%{to: Itest.PlasmaFramework.address(), data: Encoding.to_hex(data)})

case Encoding.to_binary(result) do
  "" ->
    :queue_not_added

  result ->
    next_exit_id = hd(ABI.TypeDecoder.decode(result, [{:uint, 256}]))
    next_exit_id &&& (1 <<< 160) - 1
end 

result is consistently a binary (on a clique geth node): <<8, 195, 121, 160, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 32, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 14, 81, 117, 101, 117, 101, 32, 105, 115, 32, 101, 109, 112, 116, 121, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0>>

which is decoded to 3963877391197344453575983046348115674221700746820753546331534351508065746944 and the biwtise operation turns it into:
next_exit_id &&& (1 <<< 160) - 1
0

looking at the plasma contracts code... that's not what it's supposed to happen:

plasma-contracts/plasma_framework/contracts/src/framework/ExitGameController.sol

Lines 234 to 237 in ba662a9

	function getNextExit(uint256 vaultId, address token) external view returns (uint256) {
	bytes32 key = exitQueueKey(vaultId, token);
	return exitsQueues[key].getMin();
	}

plasma-contracts/plasma_framework/contracts/src/framework/utils/PriorityQueue.sol

Lines 76 to 79 in 655a792

	function getMin() external view returns (uint256) {
	require(queue.currentSize > 0, "Queue is empty");
	return queue.heapList[1];
	}

So on ethhash node the transaction is reversed with this:
require(queue.currentSize > 0, "Queue is empty");

The question is... why do contracts behave differently on different eth nodes?

[boolafish]

found some references:

• Strange results with require() in public view/pure-type functions ethereum/solidity#4840
• Incorrect return message for require() in public/external pure/view-type functions web3/web3.js#1903

[boolafish]

would adding sth like this in ethereumx help? https://github.com/ethereum/web3.js/pull/3248/files

[boolafish]

Seems like Geth actually fixed the issue:

• eth_call cannot distinguish between success and failure with require-reason ethereum/go-ethereum#19027
• internal/ethapi: return revert reason for eth_call ethereum/go-ethereum#21083

Probably check the release and update the Geth
Okay, it is in its latest release: https://github.com/ethereum/go-ethereum/releases/tag/v1.9.15