Add fantasy-land canonical type references?

[Matt Ross]

Canonical references to the described types are provided via https://github.com/fantasyland/fantasy-land/blob/master/index.js.

These can probably be added with something like this:

addMethod('map', ...);
Stream.prototype[fl/map] = Stream.prototype.map;
_[fl/map] = _.map;

The benefit of doing this would be to provide interoperability between other fantasy-land-implementing libraries (ramda, fluture, zoom, crocks, etc).

It looks like the existing types covered are:

• fantasy-land/concat
• fantasy-land/filter
• fantasy-land/map
• fantasy-land/of
• fantasy-land/chain
• fantasy-land/ap (provided by #643)

Also worth nothing:

• fantasy-land/reduce conflicts with he current 2.x version _.reduce, but the ^3.0.0-beta.5 matches. I'm not sure how to handle this.
• fantasy-land/traverse could be fulfilled by mixing some high-level functions if fantasy-land/reduce were fulfilled.
• fantasy-land/alt could possibly be interpreted to be fulfilled by _.otherwise
• fantasy-land/empty could be fulfilled simply by () => h.of(h.nil)

Would this be desirable?

[Victor Vu]

Yes, we should do this.

fantasy-land/reduce conflicts with he current 2.x version _.reduce, but the ^3.0.0-beta.5 matches. I'm not sure how to handle this.

reduce in Highland doesn't match reduce in fantasy land, even though it's very close. The type signature for reduce in Highland is

reduce :: Highland a ~> ((b, a) -> b, b) -> Highland b

Essentially, we can do the reduce operation, but we can't unwrap the value, because it's produced asynchronously.

fantasy-land/empty could be fulfilled simply by () => h.of(h.nil)

I'd prefer using h([]) instead.

I think we can also have

• fantasy-land/zero - Same as empty.
• fantasy-land/chain - Same as flatMap.

[Matt Ross]

You're right about reduce, I overlooked the final signature somehow.

empty makes sense to me. I incorrectly wrote the one I mentioned above here as well as in #643 (#643 (comment)).

I'll look at opening a PR sometime soon.