TransportManager Refactoring to fix #300

[nikor]

Encapsulates transports in TransportManager and moves the listeners outside the transport object.

One of the test in js-libp2p fails with this pull request. There is a fix for that test on this branch: https://github.com/nikor/js-libp2p/tree/js-libp2p-switch_300

[dirkmc]

I would suggest using a Map for TransportManager.transports. This will allow you to simplify some of the other code. For example getTransports() can be implemented as

return new Map([...this.transports].map(([k, v]) => [k, v.transport]))

[nikor]

I will change it to a map, and i will try to make test pass as well :)

[jacobheun]

@nikor tests seem to be okay. The commitlint is the main thing that's failing. (Code coverage is also "failing", but it's a nominal drop in percentage). When committing dont capitalize the first letter of the commit message. fix: Remove listeners from transport object should be fix: remove listeners from transport object. It's a bit silly, but it keeps the changelog consistent.

Since this is going to be a breaking change, I think we should take advantage of the change to naming transport related things. Right now we have things like switch.transport.transports[key].transport, which makes it pretty easy to get lost. Some quick thoughts on possible changes:

Use switch.transportManager instead of switch.transport for easier reading.
switch.transportManager.get(key) and maybe privatize the transportManager._transports map
switch.transportManager.getAll() instead of getTransports

Thanks for submitting a PR for this!

[nikor]

@jacobheun Thank you for explaining why commitlint failed. I will stick to lowercase in the future :)
I will make a new commit with the naming changes.

[nikor]

Okay, i think the api changes is done.
I have make the changes need to js-libp2p on this branch https://github.com/nikor/js-libp2p/tree/js-libp2p-switch_300

[daviddias]

@jacobheun with the refactor, these need to be remade or closed. I'll let you do the review