Hey again, @nusohiro. 👋

Would you say this has considerable advantages over using pick with Array#map?

users.map(user => _.pick(user, ['name', 'age']))

edit: Oh, just realized it's not quite equivalent behavior. Is there a reason you would want an array of tuples instead of objects?

users.map(user => _.pick(user, ['name', 'age']))
// => { name, age }[]

_.pluck(users, ['name', 'age'])
// => [name, age][]

One of Radashi's core values is to avoid syntax sugar, as we prefer codebases use idiomatic JavaScript as long as it's not excessively verbose or hard to read. Basically, an over-abundance of abstractions makes a codebase harder to follow, so we don't want to encourage that.

@aleclarson

Sorry for the delay, and thanks for the thoughtful feedback!

The motivation behind this pluck function comes from a few points:

• Tuple-style output: Sometimes, especially when working with tabular data (e.g., for exporting CSVs or feeding into charting libraries), the [name, age][] format is more convenient and easier to work with than { name, age }[].
• Familiarity from Ruby: For teams with Ruby/Rails backgrounds, pluck is a well-known, concise pattern. I referenced ActiveRecord#pluck when designing this, aiming for similar utility.

Thanks for the clarification! I think we can accept this function into Radashi.

Two more questions:

• Any objection to having the keys be an array instead of rest arguments?
• What's the use case for a single key returning an array of values, rather than an array of tuples? [1, 2, 3] vs [[1], [2], [3]] – Wouldn't it be better to always return an array of tuples?

@aleclarson

Thanks, that's great to hear!

Any objection to having the keys be an array instead of rest arguments?

No strong objection at all — I'm happy to switch to an array for the keys parameter.
Let me know if you'd like me to update the PR accordingly.

What's the use case for a single key returning an array of values, rather than an array of tuples?

Good question! The main reason for returning a flat array when a single key is passed is simply because it's often easier to work with. When you're extracting something like an id, [1, 2, 3] is more convenient than [[1], [2], [3]] in most cases.
This also follows the behavior of Ruby's pluck, which returns a flat array when only one column is specified.

That said, I can see the argument for always returning a tuple-like structure for consistency. If you prefer that (i.e. always returning [[1], [2], [3]] even for one key), I’m happy to update the implementation — just let me know which direction you’d like to go!

Benchmark Results

Performance regressions of 30% or more should be investigated, unless they were anticipated. Smaller regressions may be due to normal variability, as we don't use dedicated CI infrastructure.

Thanks @nusohiro! Hope to see more from you in the future.

As you probably saw, I made the necessary changes. :)

Excuse the notification! This was released in a previous beta, so it shouldn't renotify... 😅

A stable release 12.5.0 has been published to NPM. 🚀

To install:

pnpm add radashi@12.5.0