_.chunk or _.stride function

[Soviut]

A common function I'm sure many of us run into (especially regarding pagination) is the need to break an array into chunks of a specified length. I propose a _.chunk() (or _.stride if "chunk" isn't clear enough). It would also be useful in a number of other underscore features, including my proposal for _.objects #1890.

The first value would be the list, the second value would be the chunk-size/stride-length. If the chunk size doesn't divide evenly into the length of the list, the final item should be the remainder.

_.chunk([1,2,3,4,5,6,7,8], 3);

Would yield

[[1,2,3], [4,5,6], [7,8]]

It might be worth having an optional "remainder" boolean argument that could omit the remainder if set to false, but should probably be true by default. This would be useful in cases where the chunks were going to be mapped using _.object and dangling arguments weren't desirable.

_.chunk([1,2,3,4,5,6,7,8], 3, false);

Would yield

[[1,2,3], [4,5,6]]

[jdalton]

In my implementation I opt'ed for not supporting a "remainder" function as Underscore and the like have sided on not destroying data when possible.

[Soviut]

Seems reasonable. I can't actually think of any situation where this would outright break anything.

[Soviut]

@jdalton Could you post a link to the function in the edge source?

[joshuacc]

FYI, we have _.chunk in contrib as well, though the implementation is slightly different. Note how padding behaves in contrib versus this proposal. Contrib's _.chunkAll is a closer match to this proposal.

[Soviut]

@joshuacc I feel like their implementation for _.chunk is inelegant, especially since they also have a _.chunkAll function. The implementation they give drops remainders by default which seems counter to the "don't destroy data" ideal that @jdalton mentioned. Futher, the need to supply "ignorable" items to the list to satisfy the length requirements feels very fragile.

Since both implementations are in contrib the need for a "remainder" option seems useful, however, it shouldn't truncate the remainder by default, in my opinion. Having it as an optional boolean also means it would require less code to switch between modes in conditional situations; Requiring only an expression versus being wrapped in an if statement.

[jdalton]

@jdalton Could you post a link to the function in the edge source?

See my comment on issue #1890.

Not dropping remainders by default also allows adding an option in at a later time if the request is popular enough.

[joshuacc]

@Soviut Yeah. I don't have any objections to keeping remainders by default. Just wanted to make sure folks in this thread were aware of contrib's current implementation when discussing the proposal.

If this makes it into Underscore, then we'll revise contrib to play nicely with it. 😄

[michaelficarra]

See #696. I don't think this will be accepted, and if it was I'd much rather see unfoldr instead.