Add an IRC message handling framework

[aji]

Right now a single match is used, which is good enough for the moment. However, a more robust framework for registering IRC command handlers should be implemented.

For the framework to be considered complete, it must meet the following requirements:

• A handler can exist in any module. Due to the limitations of Rust, we cannot do this in an "implied" way, like could be done with decorators in Python or annotations and component scans in Java, but it should still be possible for a closure or function defined in some distant module to be added as a message handler.
• Multiple handlers can be registered for the same IRC message, but not given any sort of priority over each other. The lack of priority ensures predictable message handling behavior.
• A handler should be able to alter the Evented the message arrived on, and alter the global Reactor. The Evented portion of this should seem obvious, but the event loop part is a bit more subtle. We may want to spawn or close a connection while in a message handler, so it's important to be able to alter the Reactor as well.

[BlacklightShining]

My immediate response to this is to a field tocore::Bot to store handlers (a HashMap<Vec<u8>, Vec<Handler>>?) and provide some kind of interface to let other modules add (and remove?) them. And the last requirement seems pretty easy; Handlers could have Fns that take mutable references to the EventedCtl and Reactor.

[aji]

yeah, passing a mutable reference to the EventedCtl is possible, but only because Bot has both &mut self and &mut EventedCtl as arguments to on_line. the sneaky problem with this (and part of why EventedCtl etc. exists) is because when one struct is inside of another, you can't pass mutable references to both without upsetting the borrow checker. that's why Bot is not inside of EventedCtl, and vice versa.

[BlacklightShining]

…hmm, the Reactor is local to Bot::run(). I think that's gonna have to change if we want to lend it to handlers…could we make it a Bot field, too?

[aji]

this is blocking all the interesting work, so I'll try to get to it this weekend

[aji]

this is sort of handled in the nonsense here now https://github.com/aji/miau/blob/master/src/network.rs#L110-L142

miles ahead of what we had before