perf: various performance improvements

[moonlightaria]

Prerequisites checklist

• [X ] I have read the contributing guidelines.

What is the purpose of this pull request? (put an "X" next to an item)

[ ] Documentation update
[ ] Bug fix (template)
[ ] New rule (template)
[ ] Changes an existing rule (template)
[ ] Add autofix to a rule
[ ] Add a CLI option
[ ] Add something to the core
[ X] Other, please explain:

What changes did you make? (Give an overview)

made various performance improvements to various parts of the code mainly based around removing unnecessary object creation and overhead from the functional tools (map, filter, reduce)

Is there anything you'd like reviewers to focus on?

[linux-foundation-easycla]

The committers listed above are authorized under a signed CLA.

• ✅ login: moonlightaria / name: moonlightaria (58a2d19, 31de562, eb9635d, 3aad578, 7f3540e, 129f9e6, cb7639b, 85ea5e6, a876b89, d85fcea, 27d2eeb, 3ea3d29)

[netlify]

✅ Deploy Preview for docs-eslint canceled.

Name	Link
🔨 Latest commit	3ea3d29
🔍 Latest deploy log	https://app.netlify.com/sites/docs-eslint/deploys/644cd0ca56143d00085191c4

[mdjermanovic]

Thanks for the PR!

For perf-related improvements, we should include before & after data that shows the improvement. Can you provide some tests and results? For rules, you could use TIMING.

[mdjermanovic]

overhead from the functional tools (map, filter, reduce)

These methods are commonly used, and therefore highly optimized in JS engines. I wouldn't expect for-loops to be noticeably faster (in fact, it might be the opposite), and that might depend on the engine and its version, so unless there's a significant difference in performance, I'm not in favor of making these changes.

[moonlightaria]

benchs.zip
here are the benchmarks on before and after the changes i just ran

for foreach it is often faster for very small collections since there is less code for js to compile but filter and map always creates a new array which becomes a performance burden expecially when chained since each on creates it own array

for some the improvement are hard to spot since it is removing garbage collection which can appear in random locations outside of the function

[nzakas]

@moonlightaria can you please paste a summary in this PR? Generally we try not to download zip files of unknown origins. :)

[moonlightaria]

benchmark command: node --cpu-prof ../../bin/eslint.js large.js --no-ignore
decided to run new benchmarks and several for better info, lost benchmark command used during initial runs so used new ones

some improvements are minimal while other are major, optimizations which reduce memory allocation remove gc cycles which can occur outside of the functions which isn't captured in this quick summery

if functions didn't appear it means no trace was grabbed during the execution of that function generally meaning that execution time to too small matter

addparensindent:
old:
438.54 ms
411.72 ms
459.63 ms
avg: 436.63 ms

new:
433.95 ms
437.45 ms
429.41 ms
avg: 433.60 ms

removeLeadingZeros:
old:
30.34 ms
20.03 ms
27.29 ms

new:
didn't appear
didn't appear
1.26 ms

removeTrailingZeros:
old:
8.04 ms
22.84 ms
11.05 ms

new:
didn't appear
didn't appear
2.61 ms

checklist:
old:
3.93 ms
.76 ms
1.24 ms
avg: 1.97 ms

new:
.42 ms
.69 ms
2.13 ms
avg: 1.08 ms

calculateStatsPerFile:
old:
26.82 ms
8.87 ms
23.44 ms
avg: 19.71 ms

new:
4.82 ms
4.88 ms
9.84 ms
avg: 6.51 ms

calculateStatsPerRun:
no appearance on any profile, same optimization as calculateStatsPerFile

checkProgramForMaxLength:
old:
168.78 ms
148.38 ms
178.24 ms
avg: 165.13 ms

new:
157.43 ms
153.75 ms
151.78 ms
avg: 154.32 ms

TokenInfo constructor:
old:
34.92 ms
39.39 ms
38.49 ms
avg: 37.6

new:
24.88 ms
34.23 ms
25.47 ms
avg: 28.19

check:
no appearance on any profile, replaced filter array creation and length check with incrementing an integer

create:
old:
37.75 ms
41.10 ms
33.16 ms
avg: 37.33 ms

new:
37.12 ms
39.41 ms
41.81 ms
avg: 39.44 ms

[github-actions]

Hi everyone, it looks like we lost track of this pull request. Please review and see what the next steps are. This pull request will auto-close in 7 days without an update.

[Rec0iL99]

benchmark command: node --cpu-prof ../../bin/eslint.js large.js --no-ignore decided to run new benchmarks and several for better info, lost benchmark command used during initial runs so used new ones

some improvements are minimal while other are major, optimizations which reduce memory allocation remove gc cycles which can occur outside of the functions which isn't captured in this quick summery

if functions didn't appear it means no trace was grabbed during the execution of that function generally meaning that execution time to too small matter

addparensindent: old: 438.54 ms 411.72 ms 459.63 ms avg: 436.63 ms

new: 433.95 ms 437.45 ms 429.41 ms avg: 433.60 ms

removeLeadingZeros: old: 30.34 ms 20.03 ms 27.29 ms

new: didn't appear didn't appear 1.26 ms

removeTrailingZeros: old: 8.04 ms 22.84 ms 11.05 ms

new: didn't appear didn't appear 2.61 ms

checklist: old: 3.93 ms .76 ms 1.24 ms avg: 1.97 ms

new: .42 ms .69 ms 2.13 ms avg: 1.08 ms

calculateStatsPerFile: old: 26.82 ms 8.87 ms 23.44 ms avg: 19.71 ms

new: 4.82 ms 4.88 ms 9.84 ms avg: 6.51 ms

calculateStatsPerRun: no appearance on any profile, same optimization as calculateStatsPerFile

checkProgramForMaxLength: old: 168.78 ms 148.38 ms 178.24 ms avg: 165.13 ms

new: 157.43 ms 153.75 ms 151.78 ms avg: 154.32 ms

TokenInfo constructor: old: 34.92 ms 39.39 ms 38.49 ms avg: 37.6

new: 24.88 ms 34.23 ms 25.47 ms avg: 28.19

check: no appearance on any profile, replaced filter array creation and length check with incrementing an integer

create: old: 37.75 ms 41.10 ms 33.16 ms avg: 37.33 ms

new: 37.12 ms 39.41 ms 41.81 ms avg: 39.44 ms

@moonlightaria has provided some form of summary that was requested.

cc @mdjermanovic

[github-actions]

Hi everyone, it looks like we lost track of this pull request. Please review and see what the next steps are. This pull request will auto-close in 7 days without an update.

[Rec0iL99]

benchmark command: node --cpu-prof ../../bin/eslint.js large.js --no-ignore decided to run new benchmarks and several for better info, lost benchmark command used during initial runs so used new ones
some improvements are minimal while other are major, optimizations which reduce memory allocation remove gc cycles which can occur outside of the functions which isn't captured in this quick summery
if functions didn't appear it means no trace was grabbed during the execution of that function generally meaning that execution time to too small matter
addparensindent: old: 438.54 ms 411.72 ms 459.63 ms avg: 436.63 ms
new: 433.95 ms 437.45 ms 429.41 ms avg: 433.60 ms
removeLeadingZeros: old: 30.34 ms 20.03 ms 27.29 ms
new: didn't appear didn't appear 1.26 ms
removeTrailingZeros: old: 8.04 ms 22.84 ms 11.05 ms
new: didn't appear didn't appear 2.61 ms
checklist: old: 3.93 ms .76 ms 1.24 ms avg: 1.97 ms
new: .42 ms .69 ms 2.13 ms avg: 1.08 ms
calculateStatsPerFile: old: 26.82 ms 8.87 ms 23.44 ms avg: 19.71 ms
new: 4.82 ms 4.88 ms 9.84 ms avg: 6.51 ms
calculateStatsPerRun: no appearance on any profile, same optimization as calculateStatsPerFile
checkProgramForMaxLength: old: 168.78 ms 148.38 ms 178.24 ms avg: 165.13 ms
new: 157.43 ms 153.75 ms 151.78 ms avg: 154.32 ms
TokenInfo constructor: old: 34.92 ms 39.39 ms 38.49 ms avg: 37.6
new: 24.88 ms 34.23 ms 25.47 ms avg: 28.19
check: no appearance on any profile, replaced filter array creation and length check with incrementing an integer
create: old: 37.75 ms 41.10 ms 33.16 ms avg: 37.33 ms
new: 37.12 ms 39.41 ms 41.81 ms avg: 39.44 ms

@moonlightaria has provided some form of summary that was requested.

cc @mdjermanovic

cc @nzakas @mdjermanovic

[github-actions]

Hi everyone, it looks like we lost track of this pull request. Please review and see what the next steps are. This pull request will auto-close in 7 days without an update.

[Rec0iL99]

Friendly ping @nzakas @mdjermanovic, this pr is open for quite a while now. I think author has provided the requested info. How do we proceed?

[nzakas]

Thanks, sorry for losing track of this. I feel comfortable with the changes, though the perf improvement seems a bit small and we are arguably losing some readability along the way.

I'm okay with merging if @mdjermanovic approves; also okay with not merging.

[mdjermanovic]

LGTM, thanks!

[nzakas]

@moonlightaria we'd like to pay you for this contribution. Please contact us at contact (at) eslint (dot) org.