yarn add --dev eslint eslint-config-ssense
Because of the current inability for sharable configs to supply their dependencies you will also need to:
yarn add --dev \
babel-eslint \
eslint-config-airbnb-base \
eslint-config-prettier \
eslint-plugin-import \
eslint-plugin-fp \
eslint-plugin-flowtype
Edit your package.json
For client-side projects:
"eslintConfig": {
"extends": "ssense/client"
}
/client
specializations are that it permits browser globals.
For server-side projects:
"eslintConfig": {
"extends": "ssense/server"
}
/server
specializations are that it permits node
globals.
For general projects (or also server-side) you can use the root config which is the same as /server
:
"eslintConfig": {
"extends": "ssense"
}
airbnb-base
↗
We extend the AirBnB rules for historical reasons. Our configuration will continue to evolve and may not be based on it one day if we eventually disable or adjust too much of it via overrides.
prettier
↗
We disable all stylistic rules that prettier takes care of for us.
flowtype
↗
We enforce static typing at SSENSE and so extend flowtype eslint rules that help devlopers use flow.
fp
↗
Provides rules that help enforce functional programming.
Talks around FP
-
The Value of Values – Rich Hickey
-
Are we there yet? – Rich Hickey
-
The Functional Final Frontier – David Nolen
-
Immutability, interactivity & JavaScript – David Nolen
Writings around FP
import
↗
Provides rules that help prevent import bugs and enforces style.
This section contains documentation about certain (not all) rules we enforce. Each rule section contains rationale and pass/fail examples. Over time we will complete exhaustive documentation. So far we have focused on significant deviations from our AirBnB inheritance.
We do not use semicolons because omitting them reduces visual noise, and so our code is more legible. Also, for writing, a day of coding with semicolons wears more on the fingers/hand than code without.
Further reading
Fail
it("foobar", () => {
assert(1, foo(1));
});
Pass
it("foobar", () => {
assert(1, foo(1))
})
We use double quotes because it is more consistent with other languages. For example some treat single/double as different types (Java, Haskell, PureScript, …), don’t even have single quotes (Clojure), or idiomatically use double (HTML). It is therefore better (assuming a polyglot programmer) for habit building and retention to use double quotes as well in JavaScript.
Fail
import Foo from 'Foo'
console.log('Foo is: %j', Foo)
Pass
import Foo from "Foo"
console.log("Foo is: %j", Foo)
Up to three allowed. Two empty lines are not enough to clearly partition major sections of a module.
Fail
import Foo from "Foo"
Foo.bar()
Pass
import Foo from "Foo"
Foo.bar()
Instead of relying on ad-hoc namespaces we should always write modules that support using default
for this functionality; that is consumers being able to do either of:
import F from "ramda"
import { compose, filter } from "ramda"
-
This is more like CommonJS which makes transition from
require
easier. -
This is simpler for developers because they have fewer options.
-
This is easier to read;
* as …
scattered multiple times throughout imports is noisy.
Fail
import * as Foo from "Foo"
Pass
import Foo from "Foo"
We use import
syntax so no need for require
anymore.
Fail
const F = require("ramda")
Pass
import F from "ramda"
Functional programming works better with known and explicit parameters. Also, having an undefined number of parameters does not work well with currying.
Fail
const sum = () => {
const numbers = Array.prototype.slice.call(arguments)
return numbers.reduce((a, b) => a + b)
}
sum(1, 2, 3)
Pass
const sum (numbers) => (
numbers.reduce((a, b) => a + b)
)
sum([1, 2, 3])
const args = node.arguments
Classes are nice tools to use when programming with the object-oriented paradigm, as they hold internal state and give access to methods on the instances. In functional programming, having stateful objects is more harmful than helpful, and should be replaced by the use of pure functions.
Further reading: Not Awesome: ES6 Classes; A curated list of resources on why ES6 (aka ES2015) classes are NOT awesome
Fail
class Polygon {
constructor (height, width) {
this.height = height
this.width = width
}
}
Pass
const polygon = (height, width) => ({
height: height,
width: width,
})
delete is an operator to remove fields from an object or elements from an array. This purposely mutates data, which is not wanted when doing functional programming.
Further reading: Avoid using delete operator
Fail
delete foo
delete foo.bar
delete foo[bar]
Pass
import F from "ramda"
const fooWithoutBar = F.omit(["bar"], foo)
const fooWithoutField = F.omit([bar], foo)
The use of EventEmitter with the events module provided by Node.js promotes implicit side-effects by emitting and listening to events. Instead of events, you should prefer activating the wanted effects by calling the functions you wish to use explicitly.
Probably what you should do is use a functional reactive programming library: most
, rxjs
.
Fail
import EventEmitter from "events"
Fail
const person = {
name: 'Some Name',
get age () {
return this._age
},
set age (n) {
if (n < 0) {
this._age = 0
} else if (n > 100) {
this._age = 100
} else {
this._age = n
}
}: 20
};
person.__defineGetter__("name", function () {
return this.name || "John Doe";
})
person.__defineSetter__("name", function (name) {
this.name = name.trim();
})
Pass
import F from "ramda"
const person = {
name: "Some Name",
age: 20,
}
const clamp = (n, min, max) => (
n <= min ? min :
n >= max ? max :
n
)
const setAge = (age, person) => (
F.merge(person, { age: clamp(age, 0, 100) })
)
If you want to program as if your variables are immutable, part of the answer is to not allow your variables to be reassigned. By not allowing the use of let and var, variables that you declared may not be reassigned.
Fail
let a = 1
let b = 2,
c = 3
let d
Pass
const a = 1
const b = 2,
c = 3
Loops, such as for or while loops, work well when using a procedural paradigm. In functional programming, recursion or implementation agnostic operations like map, filter and reduce are preferred.
Fail
const result = []
const elements = [1, 2, 3]
for (let i = 0; i < elements.length; i++) {
if (elements[i] > 2) {
result.push(elements[i])
}
}
for (element in elements) {
result.push(element * 10)
}
while (n < 100) {
result.push(n)
n *= 2
}
Pass
const xs = [1, 2, 3]
xs.filter((x) => (
x > 2
))
xs.map((x) => (
x * 10
))
const doubleBubble (n) => (
n >= 100
? []
: [n].concat(doubleBubble(n * 2))
)
When doing functional programming, you want to avoid having stateful objects and instead use simple JavaScript objects.
Also, this
actively thwarts function composition and functions-as-values (e.g. arguments to higher order functions) because when executed they would lose their this
context. The canonical solution would be .bind
but that burdens the programmer and degrades readability.
Fail
const object = {
numbers: [1, 2, 3],
sum: () => (
this.numbers.reduce((a, b) => a + b, 0)
),
}
object.sum()
Pass
Avoid this
so that function composition and functions-as-values works.
const object = {
numbers: [1, 2, 3],
sum: () => (
object.numbers.reduce((a, b) => a + b, 0)
),
}
Or better, think functionally, separating general functions from data.
const sum = (numbers) => (
numbers.reduce((a, b) => a + b, 0)
)
sum([1, 2, 3])