Implement Clique PoA consensus engine

[jangko]

Kovan, Rinkeby and Goerli are using this engine

[jangko]

we have to push this forward.
after examine the hive/rpc category, some of the test cases depend on PoA to quickly mine a random(?) block.

the way hive/rpc test cases constructed makes this feature become more important.
each of hive/rpc test cases are not consist of single rpc call, but a series of rpc calls.

because of this dynamic/algorithmic test nature, we need Clique implementatation even if not complete, it should able to mine a block. and then we can use it to test other part of rpc.

see https://github.com/ethereum/EIPs/blob/master/EIPS/eip-225.md

[jlokier]

It seems there are really two problems: Validating PoA, and mining PoA. It is certainly possible to do the former without the latter, the same way we validate PoW without mining PoW currently. Does hive/rpc use both of them?

[jangko]

yes, hive/rpc require both of them

[jlokier]

Is the mining PoA functionality only used for passing Hive tests, and is the code for mining large or is it trivial?

As we discussed last week, there is no need to pass Hive tests for functionality that Nimbus Eth1 doesn't intend to provide to end users - i.e. mining. For these it's better to say "we decided these Hive tests are not applicable to Nimbus Eth1". And actually it's not really useful to write large amounts of code just to pass tests for functionality that will never be used.

We do need block construction for the eth2 merge. But that doesn't use any Clique/PoA logic.

[jangko]

@mjfh, what is the status of this issue, can we close it and start using clique-code in RPC?

[mjfh]

Mostly done.

Did some statistics, fixed/tuned some unnecessary time consuming actions, and polished the API.

Should be open for review later this afternoon.

[jangko]

fixed by #815