Small update from my side. The last 5 days I have spent implementing the SN logic into the mining application itself and I think it's time to press the delete button and implement the SN logic into the core client itself.

Problems, that came up:
- Getting a stable and high available JSON RPC server running, I have one based on libevent but its highly experimental and well unter-tested
- With this approach we have additional overhead regarding the communication between the SN program and the core client in terms of signing transactions, verifying them, etc. This is one additional point of failure.
- Also, the SN would run on an extra port causing more overhead to SN operators since they have to open yet another port in their firewalls.

So, why reinvent the wheel when SN run the core client anyway and have all networking stuff on bord already:
I am right now doing the SN logic in the core client itself. Here, we have the entire networking part (incl. blacklisting) already running and well tested. All signing functions are already there and the only thing that we need is the uplink to the mining application for the code evaluation itself.

Good news: For massive testing, it's enough to have a pseudo SN, who does some very light checks on the work (code correctness, WCET, ...). This will be done today and if you like you can start right into the testnet this evening.


PS: Whoever likes to have a libevent based JSON-RPC server written in C for his own project, just tell me. You can have the SN stuff that was done in the last 5 days.

When we update the core client to do the real work checks on the SN side, this will not cause any soft/hardfork. The "SN only" have to update their clients themselves.