benchmark: add assert.deep[Strict]Equal benchmarks

[joyeecheung]

• Move numbers into configuration
• Add buffer comparison benchmark
• Add assert.deepStrictEqual benchmarks

Refs: #10282 (review)

Checklist

• make -j4 test (UNIX), or vcbuild test (Windows) passes
• tests and/or benchmarks are included
• commit message follows commit guidelines

Affected core subsystem(s)

benchmark, assert

[jasnell]

Almost there

[jasnell]

In other benchmarks we've used the pattern:

switch (conf.strict) {
  case 'strict':
    /** do stuff **/
  case 'nonstrict':
    /** do stuff **/
  default:
    throw new Error('Unsupported method');
}

This makes it easier to plug in new choices later and makes a code a bit easier to follow

[jasnell]

It's also possible to simplify this further by doing something like:

var method;
switch (conf.strict) {
  case 'strict':
    method = assert.deepStrictEqual;
    break;
  case 'nonstrict':
    method = assert.deepEqual;
    break;
  default:
    throw new Error('...');
}
bench.start();
for (i = 0; i < n; ++i)
  method(actual, expected);
bench.end(n);

Or even:

const methods = {
  'strict': assert.deepStrictEqual,
  'nonstrict': assert.deepEqual
};
const method = methods[conf.strict];
if (!method) throw new Error('...');
bench.start();
for (i = 0; i < n; ++i)
  method(actual, expected);
bench.end(n);

[joyeecheung]

@jasnell Thanks for the review, updated, PTAL.

[joyeecheung]

Aside: one of the try runs of Float32Array benchmarks raises the assertion, could be a V8 bug but I am not able to reproduce it.

[jasnell]

hmm... I'm not sure I'm seeing that. Worth pinging @nodejs/v8 tho just in case.

[joyeecheung]

FYI: I was playing with different n and len in deepequal-typedarrays.js when the assertion was raised. Can't remember what they were but definitely not 1 and 1e6 as they are now. At least n is not 1. And I don't even know if it's the old deepEqual (or deepStrictEqual) or if it's the new one in #10282 that raised this. Sorry I don't have more clues :/ I closed the session too soon, so probably not really worth digging into.

[joyeecheung]

CI: https://ci.nodejs.org/job/node-test-pull-request/6177/

[joyeecheung]

Going to land this in 24 hours

[joyeecheung]

Landed in 5e4545e

[jasnell]

would need a backport PR to land on v4