native-buffer-browserify

Node.js Buffer API, for the browser

buffer (native-buffer-browserify)

The buffer module from node.js, for the browser.

When you require('buffer') or use the Buffer global in browserify, this module will automatically load.

If you want to manually install it for some reason, do:

npm install buffer
  • Backed by Typed Arrays (Uint8Array and ArrayBuffer) (not Object, so it's fast)
  • Small bundle size (35KB) (half the size of buffer-browserify)
  • Excellent browser support (IE 6+, Chrome 4+, Firefox 3+, Safari 5.1+, Opera 11+, iOS, etc.).
  • Preserves Node API exactly.
  • Faster pretty much across the board (see perf results below)
  • .slice() returns instances of the same type (Buffer)
  • Square-bracket buf[4] notation works, even in old browsers like IE6!
  • Does not modify any browser prototypes.
  • Comprehensive test suite.
var Buffer = require('buffer/').Buffer  // use the npm module, not the core module!

The goal is to provide a Buffer API that is 100% identical to node's Buffer API. Read the official docs for a full list of supported methods.

The Buffer constructor returns a Uint8Array (as discussed above) for performance reasons, so instanceof Buffer won't work. In node Buffer.isBuffer just does instanceof Buffer, but in browserify we use a Buffer.isBuffer shim that detects our special Uint8Array-based Buffers.

If the browser is using the Typed Array implementation then modifying a buffer created by slice() will modify the original memory, just like in Node. But for the Object implementation (used in unsupported browsers), this is not possible. Therefore, do not rely on this behavior until browser support gets better. (Note: currently even Firefox isn't using the Typed Array implementation because of this bug.)

The Buffer constructor returns instances of Uint8Array that are augmented with function properties for all the Buffer API functions. We use Uint8Array so that square bracket notation works as expected -- it returns a single octet.

By augmenting the instances, we can avoid modifying the Uint8Array prototype.

See perf tests in /perf.

# Chrome 33
 
NewBuffer#bracket-notation x 11,194,815 ops/sec ±1.73% (64 runs sampled)
OldBuffer#bracket-notation x 9,546,694 ops/sec ±0.76% (67 runs sampled)
Fastest is NewBuffer#bracket-notation
 
NewBuffer#concat x 949,714 ops/sec ±2.48% (63 runs sampled)
OldBuffer#concat x 634,906 ops/sec ±0.42% (68 runs sampled)
Fastest is NewBuffer#concat
 
NewBuffer#copy x 15,436,458 ops/sec ±1.74% (67 runs sampled)
OldBuffer#copy x 3,990,346 ops/sec ±0.42% (68 runs sampled)
Fastest is NewBuffer#copy
 
NewBuffer#readDoubleBE x 1,132,954 ops/sec ±2.36% (65 runs sampled)
OldBuffer#readDoubleBE x 846,337 ops/sec ±0.58% (68 runs sampled)
Fastest is NewBuffer#readDoubleBE
 
NewBuffer#new x 1,419,300 ops/sec ±3.50% (66 runs sampled)
Uint8Array#new x 3,898,573 ops/sec ±0.88% (67 runs sampled) (used internally by NewBuffer)
OldBuffer#new x 2,284,568 ops/sec ±0.57% (67 runs sampled)
Fastest is Uint8Array#new
 
NewBuffer#readFloatBE x 1,203,763 ops/sec ±1.81% (68 runs sampled)
OldBuffer#readFloatBE x 954,923 ops/sec ±0.66% (70 runs sampled)
Fastest is NewBuffer#readFloatBE
 
NewBuffer#readUInt32LE x 750,341 ops/sec ±1.70% (66 runs sampled)
OldBuffer#readUInt32LE x 1,408,478 ops/sec ±0.60% (68 runs sampled)
Fastest is OldBuffer#readUInt32LE
 
NewBuffer#slice x 1,802,870 ops/sec ±1.87% (64 runs sampled)
OldBuffer#slice x 1,725,928 ops/sec ±0.74% (68 runs sampled)
Fastest is NewBuffer#slice
 
NewBuffer#writeFloatBE x 830,407 ops/sec ±3.09% (66 runs sampled)
OldBuffer#writeFloatBE x 508,446 ops/sec ±0.49% (69 runs sampled)
Fastest is NewBuffer#writeFloatBE
 
# Node 0.11
 
NewBuffer#bracket-notation x 10,912,085 ops/sec ±0.89% (92 runs sampled)
OldBuffer#bracket-notation x 9,051,638 ops/sec ±0.84% (92 runs sampled)
Buffer#bracket-notation x 10,721,608 ops/sec ±0.63% (91 runs sampled)
Fastest is NewBuffer#bracket-notation
 
NewBuffer#concat x 1,438,825 ops/sec ±1.80% (91 runs sampled)
OldBuffer#concat x 888,614 ops/sec ±2.09% (93 runs sampled)
Buffer#concat x 1,832,307 ops/sec ±1.20% (90 runs sampled)
Fastest is Buffer#concat
 
NewBuffer#copy x 5,987,167 ops/sec ±0.85% (94 runs sampled)
OldBuffer#copy x 3,892,165 ops/sec ±1.28% (93 runs sampled)
Buffer#copy x 11,208,889 ops/sec ±0.76% (91 runs sampled)
Fastest is Buffer#copy
 
NewBuffer#readDoubleBE x 1,057,233 ops/sec ±1.28% (88 runs sampled)
OldBuffer#readDoubleBE x 4,094 ops/sec ±1.09% (86 runs sampled)
Buffer#readDoubleBE x 1,587,308 ops/sec ±0.87% (84 runs sampled)
Fastest is Buffer#readDoubleBE
 
NewBuffer#new x 739,791 ops/sec ±0.89% (89 runs sampled)
Uint8Array#new x 2,745,243 ops/sec ±0.95% (91 runs sampled)
OldBuffer#new x 2,604,537 ops/sec ±0.93% (88 runs sampled)
Buffer#new x 1,836,218 ops/sec ±0.74% (92 runs sampled)
Fastest is Uint8Array#new
 
NewBuffer#readFloatBE x 1,111,263 ops/sec ±0.41% (97 runs sampled)
OldBuffer#readFloatBE x 4,026 ops/sec ±1.24% (90 runs sampled)
Buffer#readFloatBE x 1,611,800 ops/sec ±0.58% (96 runs sampled)
Fastest is Buffer#readFloatBE
 
NewBuffer#readUInt32LE x 502,024 ops/sec ±0.59% (94 runs sampled)
OldBuffer#readUInt32LE x 1,259,028 ops/sec ±0.79% (87 runs sampled)
Buffer#readUInt32LE x 2,778,635 ops/sec ±0.46% (97 runs sampled)
Fastest is Buffer#readUInt32LE
 
NewBuffer#slice x 1,174,908 ops/sec ±1.47% (89 runs sampled)
OldBuffer#slice x 2,396,302 ops/sec ±4.36% (86 runs sampled)
Buffer#slice x 2,994,029 ops/sec ±0.79% (89 runs sampled)
Fastest is Buffer#slice
 
NewBuffer#writeFloatBE x 721,081 ops/sec ±1.10% (86 runs sampled)
OldBuffer#writeFloatBE x 4,020 ops/sec ±1.04% (92 runs sampled)
Buffer#writeFloatBE x 1,811,134 ops/sec ±0.67% (91 runs sampled)
Fastest is Buffer#writeFloatBE

This was originally forked from buffer-browserify.

MIT. Copyright (C) Feross Aboukhadijeh, Romain Beauxis, and other contributors.