After the recent changes in ReasonML there really isn't a valid reason to keep working on this project anymore as ReasonML does a much better job, provides more safety, and in my opinion has an even nicer syntax and developer exerience.
Why use yet another sugar for JavaScript? Here are some features that go beyond a nice sytax:
- Immutable data
- Pattern matching in module functions
- No support for classes or OO forces a functional (yet pragmatic) coding style
- Think in data transformations with the pipe operator
|>
(like Unix pipes, F# & Elixir) - Standard Lib that utilizes Lodash (patched to be immutable)
- **CLI for compiling and creating projects **
- Compile time type inference checking using Flow
RedScript was created to provide a first class functional experience in the browser. I was tired of trying to coerce JavaScript into being a functional language (by not using the bad parts by convention and using other libs).
It is inspired from Elixir but it does not have all of the features (ElixirScript aims to do this). Our main goal is to provide easy interoperability with other JavaScript libraries (like CoffeeScript, not like Elm) while still providing a first class functional experience in JavaScript.
# 2.0 is a work in progress, not all features implemented
# Note, this compiler is just a prototype and is not production ready
sudo npm install -g redscript
redscript watch [fileName fileName2]
# use functions like unix pipe
"hello world " |> String.trim |> String.uppercase
#>>> "HELLO WORLD"
["foo", "bar", " baz"]
|> Enum.map((x) -> String.upcase(x))
|> inspect()
|> take(2)
# inspect-output ["FOO", "BAR", "BAZ"]
# >>> ["FOO", "BAR"]
Mutable data in JavaScript can lead to tricky bugs
state = {bar: 3}
# throws an error
state.bar = 10
# create copy and merge in new key/values
new_state = {state <- foo: 2, baz: 4}
# >> {foo: 2, bar: 3, baz: 4}
list = [1, 2]
list_two = [5, 6]
combined_list = [list <- 3, 4, list_two]
import {some_js_function} from "some-js-or-redscript-module"
defmodule PhoneUtils do
def normalize_number(number) do
return number
|> imported_func('-')
|> remove_char(' ')
|> remove_us_code
end
def remove_us_code(str) do
return str.replace(/^+1/g, '')
end
defp remove_char(str, char) do
return str.replace(/\s/g, '')
end
end
Just snake case the API you already know. To keep things tidy, functions are placed in namespaces. Collections are in the Enum namespace, Array in List, Strings are in a String namespace, etc... (however you could still just import lo-dash ES6 style to remove the namespace).
# or use pipe operator with lo-dash
List.take(["a", "b", "c"], 2) ["a", "b", "c"] |> List.take(2)
#>> ["a", "b"]
List.flatten_deep([1, [2, 3, [4]]]) [1, [2, 3, [4]]] |> List.flatten_deep
#>> [1, 2, 3, 4]
Enum.reject([1, 2, 3, 4], (n) => n % 2 == 0) [1, 2, 3, 4] |> Enum.reject((n) => n % 2 == 0)
#>> [1, 3]
Pattern matching can eliminate the use of if statements and can really clean up code. This is on the backlog but PRs are welcome!
defmodule MyUtils do
def count([]) do
return 0
end
def count([head|tail]) do
return 1 + count(tail)
end
end