benchmark should be executed in "strict mode"

[JungMinu]

Files in benchmark are not using strict mode
Strict mode makes it easier to write "secure" JavaScript.
Therefore, benchmark should be executed in strict mode

[silverwind]

+1. Given that most code runs in strict mode nowadays, I think we should benchmark in it too.

[JungMinu]

I will submit a PR soon 😄

[fl0w]

What's the difference in just running benchmarks with --use_strict rather? I imagine benchmarking can be done in both modes?

[JungMinu]

@fl0w just running benchmarks with --use_strict will fail because there are lots of variables that are not declared
For instance, in https://github.com/nodejs/node/blob/master/benchmark/idle_clients.js
net = require('net');
Therefore, a PR is necessary

[silverwind]

We don't even lint code in benchmark yet. Running

node tools/eslint/bin/eslint.js benchmark --rulesdir tools/eslint-rules --reset

shows 275 linter errors in benchmark, so it'd be quite a bit of work.

(also, the require-buffer rule can be disabled through a .eslintrc in benchmark. this circular dependency issue should be only relevant in lib)

[Fishrock123]

Note: I think not all of the benchmarks even work anymore.

[JungMinu]

@Fishrock123 I agree, there are many invalid files

[r-52]

@JungMinu Looks like some older benchmarks use also deprecated methods

[Fishrock123]

Some notes on benchmarks:

• The full suite hasn't been run in probably 3 years.
• They aren't actually that accurate. - See https://github.com/nodejs/benchmarking

[evanlucas]

PR: #5336

[Trott]

Fixed in #5336