lib: refactor join to inline stuff in loop

[yashLadha]

Checklist

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

[rexagod]

This looks costly, which may be shown by a quick benchmark or simply time-ing this for a large number of iterations.

[yashLadha]

I think V8 should do the optimization to inline the stuff and remove comparison as the comparison is already provided. Will do a timing benchmark.

[devsnek]

Can we benchmark just getting rid of this? It's been a long time since v8 6.0.

[sapics]

@devsnek
I make a benchmark native/join vs util/join at #33732 with nodejs v12, Windows 10.
native/join is much faster than util/join when array is big, however util/join is faster than native/join when array is small.

Join string performance check -- array size 1000000
native/join: 5.081699997186661 ms
  util/join: 146.29249899089336 ms

Join string performance check -- array size 5
native/join: 16.52709999680519 / 100000
  util/join: 9.16829900443554 / 100000

[yashLadha]

@sapics how about if we do a conditional check and use the method based on the length of the array to join.

If length >= 100
  -> Use native impl
else
  -> Use nodejs impl

One interesting thing to note is the performance difference in smaller parts 🤔

[sapics]

@yashLadha I think that it is better performance, generally.
When I searched previously, util/join was not used much, probably.
So, there are some possibilities that most of use-cases are < 100 😄

[github-actions]

This issue/PR was marked as stalled, it will be automatically closed in 30 days. If it should remain open, please leave a comment explaining why it should remain open.