timm

What is timm?

The 'timm' npm package is a small, fast, and pure immutable collections library. It provides utilities for working with immutable data structures, making it easier to manage state in JavaScript applications.

What are timm's main functionalities?

Immutable Object Update

This feature allows you to update an object immutably. The `set` function creates a new object with the updated value, leaving the original object unchanged.

const timm = require('timm');
const obj = { a: 1, b: 2 };
const newObj = timm.set(obj, 'b', 3);
console.log(newObj); // { a: 1, b: 3 }

Deep Merge

The `merge` function allows you to deeply merge two objects, creating a new object that combines the properties of both input objects.

const timm = require('timm');
const obj1 = { a: 1, b: { c: 2 } };
const obj2 = { b: { d: 3 } };
const mergedObj = timm.merge(obj1, obj2);
console.log(mergedObj); // { a: 1, b: { c: 2, d: 3 } }

Array Push

This feature allows you to add an element to the end of an array immutably. The `addLast` function creates a new array with the added element, leaving the original array unchanged.

const timm = require('timm');
const arr = [1, 2, 3];
const newArr = timm.addLast(arr, 4);
console.log(newArr); // [1, 2, 3, 4]

Array Remove

The `removeAt` function allows you to remove an element from an array immutably. It creates a new array without the specified element, leaving the original array unchanged.

const timm = require('timm');
const arr = [1, 2, 3];
const newArr = timm.removeAt(arr, 1);
console.log(newArr); // [1, 3]

Other packages similar to timm

immutable

The 'immutable' package provides persistent immutable data structures, including List, Map, Set, and Record. It is more comprehensive and feature-rich compared to 'timm', but also larger in size and complexity.

immer

The 'immer' package allows you to work with immutable state by using a 'draft' state that you can modify directly. It is more intuitive for developers who are used to mutable operations, but it uses a different approach compared to 'timm'.

seamless-immutable

The 'seamless-immutable' package provides immutable data structures that are backwards-compatible with normal JavaScript arrays and objects. It is similar to 'timm' in terms of simplicity and ease of use, but it offers a different API.

README

timm

Immutability helpers with fast reads and acceptable writes

Installation

$ npm install --save timm

Motivation

I know, I know... the world doesn't need yet another immutability library, especially with the likes of ImmutableJS and seamless-immutable around.

And yet... I felt the urge, at least just to cover my limited needs. ImmutableJS is a solid, comprehensive and highly-performant solution, but this power comes at a price: mixing up ImmutableJS's Maps and Lists with your plain objects can cause some friction, and reading those objects (in my case, more often than writing them) isn't that convenient.

On the other side, seamless-immutable solves the "friction" problem by using plain objects and arrays, but seems to have some performance issues (at least in my benchmarks, see below).

timm's approach: use plain objects and arrays and provide simple mutation functions to handle most common operations (suggestions are welcome!). As a bonus, timm creates new objects lazily, when it confirms that the operation will mutate the input object; in other words, operations that don't modify an object always return the object itself. This alleviates pressure on the garbage collector and lets you easily check whether an object changed after an operation: merge(obj, { a: 3 }) === obj.

Important notice: timm does not freeze the objects it provides. In other words, it doesn't protect you against inadvertently modifying them in your code. I considered deeply freezing all objects with Object.freeze(), but it is really slow. Then I considered doing this only in development (like seamless-immutable), but then modifying frozen objects will silently fail in development (unless you use strict in your code), and –worse still– succeed in production. Not good. In conclusion, be careful (or send me a suggestion to work around this!).

...Oh, I almost forgot! It's tiny: just ~200 LOC and ~1.5 kB minified & compressed!

Benchmarks

I prepared an initial benchmarking tool comparing read/write speeds in four cases:

• In-place editing (mutable)
• ImmutableJS
• timm
• seamless-immutable

All four solutions are first verified for consistency (the mutable solution obviously does not pass all tests) and then benchmarked. Benchmarks cover reading and writing object attributes at different nesting levels (root level, 2 levels and 5 levels deep), merging two small objects, and replacing an object in a 1000-long array.

Feel free to run them yourself (download the repo and then npm install && npm run benchmarks). These are my results on a Windows machine for 200k iterations (Node v6.2.0):

Some conclusions from these benchmarks:

• Reads are on par with native objects/arrays and seamless-immutable, and faster than ImmutableJS (the deeper, the faster). In fact, you cannot go faster than native objects for reading!

• Writes are much slower than in-place edits, as expected, but are much faster than seamless-immutable (even in production mode), both for objects and arrays. Compared to ImmutableJS, object writes and merges are faster (the deeper, the faster), whereas array writes are way slower (not as slow as seamless-immutable, though). For timm and seamless-immutable, write times degrade linearly with array length (and probably object size), but much more slowly for ImmutableJS (logarithmically?). This is where ImmutableJS really shines.

• Hence, what I recommend (from top to bottom):

  • If you don't need immutability, well... just mutate in peace! I mean, in place
  • If you need a complete, well-tested, rock-solid library and don't mind using a non-native API for reads: use ImmutableJS
  • If you value using plain arrays/objects above other considerations, use timm
  • If your typical use cases involve much more reading than writing, use timm as well
  • If you do a lot of writes, updating items in long arrays or attributes in fat objects, use ImmutableJS

Usage

// Consume the CommonJS ES5 module (default)
import { merge } from 'timm';
const obj = timm.merge({ a: 2 }, { b: 3 });

// Consume the native ES6 module (requires bundling)
import { merge } from 'timm/lib/timm_es6';
const obj = merge({ a: 2 }, { b: 3 });

Note: you can use the ES6 module version with Webpack 2 and its tree-shaking feature to get really small bundles, including just the timm functions you use. Make sure you disable babel-plugin-transform-es2015-modules-commonjs in your babel-loader configuration. More details here.

Arrays

addLast()

Returns a new array with an appended item or items.

Usage: addLast(array: Array<any>, val: Array<any>|any): Array<any>

arr = ['a', 'b']
arr2 = addLast(arr, 'c')
// ['a', 'b', 'c']
arr2 === arr
// false
arr3 = addLast(arr, ['c', 'd'])
// ['a', 'b', 'c', 'd']

addFirst()

Returns a new array with a prepended item or items.

Usage: addFirst(array: Array<any>, val: Array<any>|any): Array<any>

arr = ['a', 'b']
arr2 = addFirst(arr, 'c')
// ['c', 'a', 'b']
arr2 === arr
// false
arr3 = addFirst(arr, ['c', 'd'])
// ['c', 'd', 'a', 'b']

insert()

Returns a new array obtained by inserting an item or items at a specified index.

Usage: insert(array: Array<any>, idx: number, val: Array<any>|any): Array<any>

arr = ['a', 'b', 'c']
arr2 = insert(arr, 1, 'd')
// ['a', 'd', 'b', 'c']
arr2 === arr
// false
insert(arr, 1, ['d', 'e'])
// ['a', 'd', 'e', 'b', 'c']

removeAt()

Returns a new array obtained by removing an item at a specified index.

Usage: removeAt(array: Array<any>, idx: number): Array<any>

arr = ['a', 'b', 'c']
arr2 = removeAt(arr, 1)
// ['a', 'c']
arr2 === arr
// false

replaceAt()

Returns a new array obtained by replacing an item at a specified index. If the provided item is the same (referentially equal to) the previous item at that position, the original array is returned.

Usage: replaceAt(array: Array<any>, idx: number, newItem: any): Array<any>

arr = ['a', 'b', 'c']
arr2 = replaceAt(arr, 1, 'd')
// ['a', 'd', 'c']
arr2 === arr
// false

// The same object is returned if there are no changes:
replaceAt(arr, 1, 'b') === arr
// true

Collections (objects and arrays)

The following types are used throughout this section

type ArrayOrObject = Array<any>|Object;
type Key = number|string;

getIn()

Returns a value from an object at a given path. Works with nested arrays and objects. If the path does not exist, it returns undefined.

Usage: getIn(obj: ?ArrayOrObject, path: Array<Key>): any

obj = { a: 1, b: 2, d: { d1: 3, d2: 4 }, e: ['a', 'b', 'c'] }
getIn(obj, ['d', 'd1'])
// 3
getIn(obj, ['e', 1])
// 'b'

set()

Returns a new object with a modified attribute. If the provided value is the same (referentially equal to) the previous value, the original object is returned.

Usage: set(obj: any, key: Key, val: any): Object

obj = { a: 1, b: 2, c: 3 }
obj2 = set(obj, 'b', 5)
// { a: 1, b: 5, c: 3 }
obj2 === obj
// false

// The same object is returned if there are no changes:
set(obj, 'b', 2) === obj
// true

setIn()

Returns a new object with a modified nested attribute.

Notes:

• If the provided value is the same (referentially equal to) the previous value, the original object is returned.
• If the path does not exist, it will be created before setting the new value.

Usage: setIn(obj: ArrayOrObject, path: Array<Key>, val: any): ArrayOrObject

obj = { a: 1, b: 2, d: { d1: 3, d2: 4 }, e: { e1: 'foo', e2: 'bar' } }
obj2 = setIn(obj, ['d', 'd1'], 4)
// { a: 1, b: 2, d: { d1: 4, d2: 4 }, e: { e1: 'foo', e2: 'bar' } }
obj2 === obj
// false
obj2.d === obj.d
// false
obj2.e === obj.e
// true

// The same object is returned if there are no changes:
obj3 = setIn(obj, ['d', 'd1'], 3)
// { a: 1, b: 2, d: { d1: 3, d2: 4 }, e: { e1: 'foo', e2: 'bar' } }
obj3 === obj
// true
obj3.d === obj.d
// true
obj3.e === obj.e
// true

// ... unknown paths create intermediate keys:
setIn({ a: 3 }, ['unknown', 'path'], 4)
// { a: 3, unknown: { path: 4 } }

updateIn()

Returns a new object with a modified nested attribute, calculated via a user-provided callback based on the current value. If the calculated value is the same (referentially equal to) the previous value, the original object is returned.

Usage: updateIn(obj: ArrayOrObject, path: Array<Key>, fnUpdate: (prevValue: any) => any): Object

obj = { a: 1, d: { d1: 3, d2: 4 } }
obj2 = updateIn(obj, ['d', 'd1'], (val) => val + 1)
// { a: 1, d: { d1: 4, d2: 4 } }
obj2 === obj
// false

// The same object is returned if there are no changes:
obj3 = updateIn(obj, ['d', 'd1'], (val) => val)
// { a: 1, d: { d1: 3, d2: 4 } }
obj3 === obj
// true

merge()

Returns a new object built as follows: the overlapping keys from the second one overwrite the corresponding entries from the first one. Similar to Object.assign(), but immutable.

Usage:

• merge(obj1: ArrayOrObject, obj2: ?ArrayOrObject): ArrayOrObject
• merge(obj1: ArrayOrObject, ...objects: Array<?ArrayOrObject>): Object

The unmodified obj1 is returned if obj2 does not provide something new to obj1, i.e. if either of the following conditions are true:

• obj2 is null or undefined
• obj2 is an object, but it is empty
• All attributes of obj2 are referentially equal to the corresponding attributes of obj

obj1 = { a: 1, b: 2, c: 3 }
obj2 = { c: 4, d: 5 }
obj3 = merge(obj1, obj2)
// { a: 1, b: 2, c: 4, d: 5 }
obj3 === obj1
// false

// The same object is returned if there are no changes:
merge(obj1, { c: 3 }) === obj1
// true

mergeIn()

Similar to merge(), but merging the value at a given nested path.

Usage:

• mergeIn(obj1: ArrayOrObject, path: Array<Key>, obj2: ArrayOrObject): ArrayOrObject
• mergeIn(obj1: ArrayOrObject, path: Array<Key>, ...objects: Array<?ArrayOrObject>): Object

obj1 = { a: 1, d: { b: { d1: 3, d2: 4 } } }
obj2 = { d3: 5 }
obj3 = mergeIn(obj1, ['d', 'b'], obj2)
// { a: 1, d: { b: { d1: 3, d2: 4, d3: 5 } } }
obj3 === obj1
// false

// The same object is returned if there are no changes:
mergeIn(obj1, ['d', 'b'], { d2: 4 }) === obj1
// true

omit()

Returns an object excluding one or several attributes.

Usage: omit(obj: Object, attrs: Array<string>|string): Object

obj = { a: 1, b: 2, c: 3, d: 4 }
omit(obj, 'a')
// { b: 2, c: 3, d: 4 }
omit(obj, ['b', 'c'])
// { a: 1, d: 4 }

// The same object is returned if there are no changes:
omit(obj, 'z') === obj1
// true

addDefaults()

Returns a new object built as follows: undefined keys in the first one are filled in with the corresponding values from the second one (even if they are null).

Usage:

• addDefaults(obj: ArrayOrObject, defaults: ArrayOrObject): Object
• addDefaults(obj: ArrayOrObject, ...defaultObjects: Array<?ArrayOrObject>): Object

obj1 = { a: 1, b: 2, c: 3 }
obj2 = { c: 4, d: 5, e: null }
obj3 = addDefaults(obj1, obj2)
// { a: 1, b: 2, c: 3, d: 5, e: null }
obj3 === obj1
// false

// The same object is returned if there are no changes:
addDefaults(obj1, { c: 4 }) === obj1
// true

MIT license

Copyright (c) Guillermo Grau Panea 2016

Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:

The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.

THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.

Changelog

1.1.2 (October 19, 2016)

• Change how Flow type definitions are exposed so they can be used with zero config.