LES Part 2

[rphmeier]

• All request handlers implemented
• Provider implemented for Client via ProvingBlockChainClient trait (without CHT stuff, see Canonical Hash Trie calculations for full node #3458)
• Simple event handlers, constructor
• Request dispatch, although responses will lead to panic for the moment

[rphmeier]

Next PR will contain:

• CHT for full node (may require DB migration)
• Response handlers
• At least part of light sync service

[rphmeier]

I swear I used git mv when moving the light module location...

[coveralls]

Changes Unknown when pulling 58ca93c on client-provider into * on master*.

[arkpar]

Move light module into ethcore crate for IPC codegen

Whys is this needed? Codegen works just fine for ethsync crate. Maybe move to ethsync instead?

[rphmeier]

@arkpar Most of the reason was actually that the implementation of Provider for Client needed access to a lot of internals which just weren't available outside of the ethcore crate.

The net primitives are also lower level than the light sync service, although that will involve an instantiation of the LightProtocol. I will be writing the sync/serve processes within the ethsync crate, but this will be in the next PR.

[rphmeier]

Thinking on it a bit more, it should be possible to have ethcore-light as its own crate again, with the addition of a trait StateAt: BlockChainClient { } and impl<T: StateAt + ?Sized> Provider for T { ... } in ethcore.

I can add that change to this PR if you'd like.

[arkpar]

I'd like to keep ethcore crate as light (pun not intended) as possible. It is becoming monstrous already. And introducing dependency on ethcore-network does not seem right. I'm fine if it happens in the next PR.

[rphmeier]

This is ready for review again.