Basic protocol support for vanilla JS objects via mori.extend

[tgriesser]

Adds basic support for extending several protocols to vanilla JS objects via mori.extend

The protocols included thusfar: IAssociative, ICloneable, ICollection, ICounted, IEncodeClojure, IEncodeJS, IEquiv, IHash, IKVReduce, ILookup, IMap, ISeq, ISeqable

Real world use case: https://github.com/tgriesser/mori-cursors

Currently, adding a protocol looks something like:

mori.extend("IAssociative", Cursor.prototype, {
  contains_key: Cursor.prototype.contains_key,
  assoc: Cursor.prototype.assoc
})

However I'm wondering whether it'd be more concise to assume the object provided as the second arg to extend implements specific protocol functions, so you could just say:

mori.extend("IAssociative", MapCursor.prototype, 'contains_key', 'assoc')

or just

mori.extend("IAssociative", MapCursor.prototype)

Also, the IEncodeClojure is currently pretty useless, but would become useful with the proposed change here (or similar): http://dev.clojure.org/jira/browse/CLJS-842

[tgriesser]

Updated the entry point as mentioned tgriesser@c46eb42#commitcomment-7496107

[eborden]

Hey guys, is this pull request scheduled for a merge at any point?

[swannodette]

@eborden it needs to be updated to be a bit more flexible.

[tgriesser]

Any specific pointers on what I could do to update it? It's been working quite well for me (aside from the js->clj issue - http://dev.clojure.org/jira/browse/CLJS-842)

[eborden]

I'm currently using tgriesser's patch and it is working quite nicely. I'm also curious what added flexibility you see a need for.

[skrat]

@swannodette it looks pretty solid to me, and given that it allows @tgriesser to implement cursors, seems good enough reason to include this PR. Looking at your recent benchmarks against immutable.js, it makes want to make the switch, but... no cursors, no go for us.

[tgriesser]

@swannodette went ahead and updated with master if you wanted to take another look.

@skrat the "cursors" I implemented in that gh repo are definitely subpar and I'm not really planning on maintaining them. So I'm not sure I'd recommend using that implementation in anything serious, but it does demonstrate a proof of concept of using custom persistent data structures from mori which is pretty neat.

[skrat]

I wonder whether a library that exposes ClojureScript's atom and Om's cursors wouldn't be a better solution, that way the maintenance problem would be solved by having a simple wrapper. The meat would be implemented in well established and actively maintained projects.

[eborden]

@skrat This PR has other uses other than implementing cursors, such as utilizing complex types in Maps and Sets.