Support canonical encoding by adding sortKeys option to encode()

[gfx]

• https://github.com/mcollina/msgpack5 has the same option sortKeys
• msgpack-lite has pull request Add support for canonical encoding kawanet/msgpack-lite#103 has a pull-request for canonical encoding

[codecov-io]

Codecov Report

Merging #64 into master will decrease coverage by <.01%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master      #64      +/-   ##
==========================================
- Coverage    96.9%   96.89%   -0.01%     
==========================================
  Files          15       15              
  Lines         905      903       -2     
  Branches      183      183              
==========================================
- Hits          877      875       -2     
  Misses         28       28

Impacted Files	Coverage Δ
src/Encoder.ts	97.78% <100%> (-0.02%)	⬇️
src/encode.ts	100% <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 aed070f...3c0ee99. Read the comment docs.

[gfx]

FYI @sergeyzenchenko

[sergeyzenchenko]

LGTM, but left one comment about sorting. Take a look at it

[sergeyzenchenko]

Maybe we should have two separate branches? One with keys array and one without? it will be better from performance perspective. But maybe it's over optimization

[gfx]

Maybe you are right because most of the cases do not set sortKeys.

[gfx]

Unfortunately, Object.keys() version is always faster.

benchmark script:

import { encode, decode } from "../src";
// @ts-ignore
import _ from "lodash";

const data: Record<string, number> = {};
for (let i = 0; i < 100; i++) {
  data["foo" + i] = i;
}

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

// run

console.time("encode sortKeys: false");
for (let i = 0; i < 100000; i++) {
  encode(data, { sortKeys: false });
}
console.timeEnd("encode sortKeys: false");

console.time("encode sortKeys: true");
for (let i = 0; i < 100000; i++) {
  encode(data, { sortKeys: true });
}
console.timeEnd("encode sortKeys: true");

result of for-in version:

$ npx ts-node benchmark/sandbox.ts
encode sortKeys: false: 2681.488ms
encode sortKeys: true: 2374.899ms

result of Object.keys() version (i.e. the same as this PR):

$ npx ts-node benchmark/sandbox.ts
encode sortKeys: false: 1336.151ms
encode sortKeys: true: 2343.010ms

I think for-in version uses less memory, but it's much slower.

So I don't change this PR.