The Essential Guide To JavaScript’s Newest Data Type: BigInt

• 8 min read
• TypeScript, JavaScript, Techniques
• Share on Twitter, LinkedIn

About The Author

Faraz is a professional JavaScript developer who is passionate about moving the web forward and promoting patterns and ideas that will make development more … More about Faraz ↬

Email Newsletter

Your (smashing) email

Weekly tips on front-end & UX.
Trusted by 200,000+ folks.

The BigInt data type aims to enable JavaScript programmers to represent integer values larger than the range supported by the Number data type. The ability to represent integers with arbitrary precision is particularly important when performing mathematical operations on large integers. With BigInt, integer overflow will no longer be an issue.

Additionally, you can safely work with high-resolution timestamps, large integer IDs, and more without having to use a workaround. BigInt is currently a stage 3 proposal. Once added to the specification, it will become the second numeric data type in JavaScript, which will bring the total number of supported data types to eight:

• Boolean
• Null
• Undefined
• Number
• BigInt
• String
• Symbol
• Object

In this article, we will take a good look at BigInt and see how it can help overcome the limitations of the Number type in JavaScript.

The Problem

The lack of an explicit integer type in JavaScript is often baffling to programmers coming from other languages. Many programming languages support multiple numeric types such as float, double, integer, and bignum, but that’s not the case with JavaScript. In JavaScript, all numbers are represented in double-precision 64-bit floating-point format as defined by the IEEE 754-2008 standard.

Under this standard, very large integers that cannot be exactly represented are automatically rounded. To be precise, the Number type in JavaScript can only safely represent integers between -9007199254740991 (-(2⁵³-1)) and 9007199254740991 (2⁵³-1). Any integer value that falls out of this range may lose precision.

This can be easily examined by executing the following code:

console.log(9999999999999999);    // → 10000000000000000

This integer is larger than the largest number JavaScript can reliably represent with the Number primitive. Therefore, it’s rounded. Unexpected rounding can compromise a program’s reliability and security. Here’s another example:

// notice the last digits
9007199254740992 === 9007199254740993;    // → true

JavaScript provides the Number.MAX_SAFE_INTEGER constant that allows you to quickly obtain the maximum safe integer in JavaScript. Similarly, you can obtain the minimum safe integer by using the Number.MIN_SAFE_INTEGER constant:

const minInt = Number.MIN_SAFE_INTEGER;

console.log(minInt);         // → -9007199254740991

console.log(minInt - 5);     // → -9007199254740996

// notice how this outputs the same value as above
console.log(minInt - 4);     // → -9007199254740996

The Solution

As a workaround to these limitations, some JavaScript developers represent large integers using the String type. The Twitter API, for example, adds a string version of IDs to objects when responding with JSON. Additionally, a number of libraries such as bignumber.js have been developed to make working with large integers easier.

With BigInt, applications no longer need a workaround or library to safely represent integers beyond Number.MAX_SAFE_INTEGER and Number.Min_SAFE_INTEGER. Arithmetic operations on large integers can now be performed in standard JavaScript without risking loss of precision. The added benefit of using a native data type over a third-party library is better run-time performance.

To create a BigInt, simply append n to the end of an integer. Compare:

console.log(9007199254740995n);    // → 9007199254740995n
console.log(9007199254740995);     // → 9007199254740996

Alternatively, you can call the BigInt() constructor:

BigInt("9007199254740995");    // → 9007199254740995n

BigInt literals can also be written in binary, octal or hexadecimal notation:

// binary
console.log(0b100000000000000000000000000000000000000000000000000011n);
// → 9007199254740995n

// hex
console.log(0x20000000000003n);
// → 9007199254740995n

// octal
console.log(0o400000000000000003n);
// → 9007199254740995n

// note that legacy octal syntax is not supported
console.log(0400000000000000003n);
// → SyntaxError

Keep in mind that you can’t use the strict equality operator to compare a BigInt to a regular number because they are not of the same type:

console.log(10n === 10);    // → false

console.log(typeof 10n);    // → bigint
console.log(typeof 10);     // → number

Instead, you can use the equality operator, which performs implicit type conversion before compering its operands:

console.log(10n == 10);    // → true

All arithmetic operators can be used on BigInts except for the unary plus (+) operator:

10n + 20n;    // → 30n
10n - 20n;    // → -10n
+10n;         // → TypeError: Cannot convert a BigInt value to a number
-10n;         // → -10n
10n * 20n;    // → 200n
20n / 10n;    // → 2n
23n % 10n;    // → 3n
10n ** 3n;    // → 1000n

let x = 10n;
++x;          // → 11n
--x;          // → 10n

The reason that the unary plus (+) operator is not supported is that some programs may rely on the invariant that + always produces a Number, or throws an exception. Changing the behavior of + would also break asm.js code.

Naturally, when used with BigInt operands, arithmetic operators are expected to return a BigInt value. Therefore, the result of the division (/) operator is automatically truncated. For example:

25 / 10;      // → 2.5
25n / 10n;    // → 2n

Implicit Type Conversion

Because implicit type conversion could lose information, mixed operations between BigInts and Numbers are not allowed. When mixing large integers and floating-point numbers, the resulting value may not be accurately representable by BigInt or Number. Consider the following example:

(9007199254740992n + 1n) + 0.5

The result of this expression is outside of the domain of both BigInt and Number. A Number with a fractional part cannot be accurately converted to a BigInt. And a BigInt larger than 2⁵³ cannot be accurately converted to a Number.

As a result of this restriction, it’s not possible to perform arithmetic operations with a mix of Number and BigInt operands. You also cannot pass a BigInt to Web APIs and built-in JavaScript functions that expect a Number. Attempting to do so will cause a TypeError:

10 + 10n;    // → TypeError
Math.max(2n, 4n, 6n);    // → TypeError

Note that relational operators do not follow this rule, as shown in this example:

10n > 5;    // → true

If you want to perform arithmetic computations with BigInt and Number, you first need to determine the domain in which the operation should be done. To do that, simply convert either of the operands by calling Number() or BigInt():

BigInt(10) + 10n;    // → 20n
// or
10 + Number(10n);    // → 20

When encountered in a Boolean context, BigInt is treated similar to Number. In other words, a BigInt is considered a truthy value as long as it’s not 0n:

if (5n) {
    // this code block will be executed
}

if (0n) {
    // but this code block won't
}

No implicit type conversion between BigInt and Number types occurs when sorting an array:

const arr = [3n, 4, 2, 1n, 0, -1n];

arr.sort();    // → [-1n, 0, 1n, 2, 3n, 4]

Bitwise operators such as |, &, <<, >>, and ^ operate on BigInts in a similar way to Numbers. Negative numbers are interpreted as infinite-length two’s complement. Mixed operands are not allowed. Here are some examples:

90 | 115;      // → 123
90n | 115n;    // → 123n
90n | 115;     // → TypeError

The BigInt Constructor

As with other primitive types, a BigInt can be created using a constructor function. The argument passed to BigInt() is automatically converted to a BigInt, if possible:

BigInt("10");    // → 10n
BigInt(10);      // → 10n
BigInt(true);    // → 1n

Data types and values that cannot be converted throw an exception:

BigInt(10.2);     // → RangeError
BigInt(null);     // → TypeError
BigInt("abc");    // → SyntaxError

You can directly perform arithmetic operations on a BigInt created using a constructor:

BigInt(10) * 10n;    // → 100n

When used as operands of the strict equality operator, BigInts created using a constructor are treated similar to regular ones:

BigInt(true) === 1n;    // → true

Library Functions

JavaScript provides two library functions for representing BigInt values as signed or unsigned integers:

• BigInt.asUintN(width, BigInt): wraps a BigInt between 0 and 2^width-1
• BigInt.asIntN(width, BigInt): wraps a BigInt between -2^width-1 and 2^width-1-1

These functions are particularly useful when performing 64-bit arithmetic operations. This way you can stay within the intended range.

Browser Support And Transpiling

At the time of this writing, Chrome +67 and Opera +54 fully support the BigInt data type. Unfortunately, Edge and Safari haven’t implemented it yet. Firefox doesn’t support BigInt by default, but it can be enabled by setting javascript.options.bigint to true in about:config. An up-to-date list of supported browsers is available on Can I use….

Unluckily, transpiling BigInt is an extremely complicated process, which incurs hefty run-time performance penalty. It’s also impossible to directly polyfill BigInt because the proposal changes the behavior of several existing operators. For now, a better alternative is to use the JSBI library, which is a pure-JavaScript implementation of the BigInt proposal.

This library provides an API that behaves exactly the same as the native BigInt. Here’s how you can use JSBI:

import JSBI from './jsbi.mjs';

const b1 = JSBI.BigInt(Number.MAX_SAFE_INTEGER);
const b2 = JSBI.BigInt('10');

const result = JSBI.add(b1, b2);

console.log(String(result));    // → '9007199254741001'

An advantage of using JSBI is that once browser support improves, you won’t need to rewrite your code. Instead, you can automatically compile your JSBI code into native BigInt code by using a babel plugin. Furthermore, the performance of JSBI is on par with native BigInt implementations. You can expect wider browser support for BigInt soon.

Conclusion

BigInt is a new data type intended for use when integer values are larger than the range supported by the Number data type. This data type allows us to safely perform arithmetic operations on large integers, represent high-resolution timestamps, use large integer IDs, and more without the need to use a library.

It’s important to keep in mind that you cannot perform arithmetic operations with a mix of Number and BigInt operands. You’ll need to determine the domain in which the operation should be done by explicitly converting either of the operands. Moreover, for compatibility reasons, you are not allowed to use the unary plus (+) operator on a BigInt.

What do you think? Do you find BigInt useful? Let us know in the comments!

Further Reading

• Converting Plain Text To Encoded HTML With Vanilla JavaScript
• Setting And Persisting Color Scheme Preferences With CSS And A “Touch” Of JavaScript
• Modern CSS Layouts: You Might Not Need A Framework For That
• Switching It Up With HTML’s Latest Control

(dm, yk, il, mrn)