Add support for eth_chainId

[halfalicious]

Fixes #4810
Implements https://github.com/ethereum/EIPs/blob/master/EIPS/eip-695.md

I added a chainId() function to the libethereum\Interface class which returns 0 by default (since chainId values start @ 0x1 with MainNet), along with the corresponding function to libethereum\ClientBase which retrieves the current chain id from BlockChain::chainParams. I then surfaced the function in the JSON-RPC API by adding it to libweb3jsonrpc\eth.json and regenerating EthFace.h via jsonrpcstub, and removing unnecessary changes from the generated file.

I tested my changes on Ubuntu with eth connected to Ropsten and a JSON-RPC server setup via the jsonrpcproxy.py script:

localadminuser@ELLIOTT:~$ curl -X POST --data '{"jsonrpc":"2.0","method":"eth_chainId","params":[],"id":1}' http://localhost:8545
{"id":1,"jsonrpc":"2.0","result":"0x3"}
localadminuser@ELLIOTT:~$ 

[codecov-io]

Codecov Report

Merging #4975 into develop will decrease coverage by 0.07%.
The diff coverage is 7.69%.

@@             Coverage Diff             @@
##           develop    #4975      +/-   ##
===========================================
- Coverage    61.08%   61.01%   -0.08%     
===========================================
  Files          350      350              
  Lines        28338    28319      -19     
  Branches      2884     2884              
===========================================
- Hits         17310    17278      -32     
- Misses       10069    10073       +4     
- Partials       959      968       +9

[chfast]

Besides single nit, it looks good to me. @gumb0 would you double check it?

[chfast]

Do not repeat virtual keyword if override is used.

[halfalicious]

Thanks for the feedback @chfast! I followed the approach used by other functions in the class e.g.:

https://github.com/ethereum/cpp-ethereum/blob/3f2ac7c4f9e1543eb4f31cbdcdf8028eb9bfe0b6/libethereum/ClientBase.h#L130

After reading more about the virtual and override keywords, it looks like neither of them are required:

• Virtual: This keyword is redundant because ClientBase::chainId() is implicitly virtual because it overrides a virtual function (Interface::chainId())
• Override: This keyword isn't required, but it can be help one avoid runtime bugs since it tells the compiler to verify that there's a function in the parent class with the same signature and to raise an error if that's not the case

As such, I think it makes sense to remove the virtual keyword. However, I can also see the value in keeping both keywords in the interest of code consistency. What are your thoughts?

[gumb0]

Yes, override is helpful and having both virtual and override is redundant. I'm fine with making it less consistent and leaving only override.
You could also remove all other virtual from other overridden methods.

[halfalicious]

Thanks @gumb0, I've removed virtual from all of the ClientBase functions which override other functions.

[gumb0]

Looks good to me

[gumb0]

It would be great if you could also document how you generate stub files with jsonrpcstub, because usually no one can't make it work.