Improve array decoding performance

[sergeyzenchenko]

Hi @gfx I've used this msgpack library in my project and found small optimization that improves decoding speed by 3-4%.

[sergeyzenchenko]

For large arrays performance difference can be even bigger and also it will produce less trash for GC

[codecov-io]

Codecov Report

Merging #32 into master will increase coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master      #32      +/-   ##
==========================================
+ Coverage   96.14%   96.14%   +<.01%     
==========================================
  Files          14       14              
  Lines         752      753       +1     
  Branches      158      158              
==========================================
+ Hits          723      724       +1     
  Misses         12       12              
  Partials       17       17

Impacted Files	Coverage Δ
src/Decoder.ts	99.27% <100%> (ø)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update c022de5...3d50366. Read the comment docs.

[gfx]

Tested with this benchmark script:

/* eslint-disable no-console */
import { encode, decode } from "../src";
// @ts-ignore
import _ from "lodash";

const data = _.cloneDeep(new Array(1000).fill([[1], [1, 2, 3], [1, 2, 3, 4, 5]]));

// warm up
const encoded = encode(data);
decode(encoded);

// run

console.time("encode");
for (let i = 0; i < 10000; i++) {
  encode(data);
}
console.timeEnd("encode");

console.time("decode");
for (let i = 0; i < 10000; i++) {
  decode(encoded);
}
console.timeEnd("decode");

And this PR significantly improves decoding:

before:

$ npx ts-node benchmark/sandbox.ts
encode: 2753.986ms
decode: 1855.263ms

after:

$ npx ts-node benchmark/sandbox.ts
encode: 2393.308ms
decode: 3804.347ms

Great!

[gfx]

(CI failed because of SauceLabs credentials; will fix it later https://docs.travis-ci.com/user/environment-variables/ )

[nelsonlaquet]

I wonder if you could improve the performance further on certain engines by allocating an array without holes in it:

array: Array.from({ length: size }),

Source:
https://v8.dev/blog/elements-kinds
https://2ality.com/2018/12/creating-arrays.html

[gfx]

Tried it, but Array.from({ length: size }) makes decoding 10x slower than new Array(size) :-(