Split the parsing and the IO

[hardbyte]

I'm convinced we should clearly split the parsing and the IO for each interface. This obviously effects the entire library - but it should all be internal changes.

See this http 1.1 example or this great pycon16 talk.

[felixdivo]

I just watched that talk you linked yesterday and now get what you meant. ^^ I initially thought about the current can.io.* package which didn't make too much sense.

I totally agree with the need for splitting the parsing and the I/O. Although I think it would mainly benefit the testing & readability of this library, rather than the reusability in other projects. But who knows.

But there are a few more problems that I personally had with understanding the code of this library, which should be part of this discussion. I think a lot of the project structure could be made quite a bit simpler to make it easier to get started. When I have some time I will open a separate Issue for that to start a discussion on that in general.

[felixdivo]

@hardbyte Is this also a discussion?