fixes #349: add _.partition function

[michaelficarra]

Fixes #349.

[jashkenas]

This one is really borderline because it's so close to the simplest use case for reduce ... but, alright.

[davidchambers]

Is it okay to use "predicate" now? The documentation doesn't currently use this term, though it should: "iterator" is both vague and slightly misleading.

I'd like to open a pull request to replace "iterator" with "predicate" where appropriate (in the code as well as the documentation). Would such a pull request be considered?

[michaelficarra]

Oh, I just used what was natural to me. I'd +1 such a PR.

[jashkenas]

Yes, that would be a great change.

[davidchambers]

#1435

[ghost]

This looks very similar to _.groupBy and has less features than the underscore-contrib version of _.partition

[michaelficarra]

@Dw40: The underscore-contrib function named partition has nothing to do with this function. The underscore-contrib partition is more like the proposed (and, in my opinion, better named) groupsOf.

[ghost]

It's still conflicting with a companion project. The way Underscore's _.partition is now doesn't hold a lot of value over what can be achieved by:

_.groupBy([0, 1, 2, 3, 4, 5], isOdd);
// { 'false': [0, 2, 4], 'true': [1, 3, 5] }

[bjmiller]

Why not change the function name in contrib, then? I suspect that Fogus would be OK with that.

[ghost]

I'm not sure there is enough to differentiate it from _.groupBy. As Jeremy said _.partition was borderline because it was so close to _.reduce, but it's even closer to _.groupBy.