refactor: use async/await instead then() in promises

[multivoltage]

Checklist

• [x] run npm run test and npm run benchmark
• [x] tests and/or benchmarks are included
• [x] documentation is changed or added
• [x] commit message and code follows the Developer's Certification of Origin and the Code of conduct

Why this PR?

I added support for html-minifier-terser in a separate PR. To to that I wrote some "await" and "async". Matteo Collina advised that it's better to have all "await" or all callback for promises. So I tried to remove .then() and .catch() in favour of await asynk

[multivoltage]

I tried to edit all calback

After that I tried npm run exaple && autocannon -c 100 -d 5 -p 10 localhost:3000

In master

┌─────────┬───────┬────────┬────────┬────────┬───────────┬─────────┬────────┐
│ Stat    │ 2.5%  │ 50%    │ 97.5%  │ 99%    │ Avg       │ Stdev   │ Max    │
├─────────┼───────┼────────┼────────┼────────┼───────────┼─────────┼────────┤
│ Latency │ 98 ms │ 101 ms │ 161 ms │ 174 ms │ 107.03 ms │ 16.1 ms │ 198 ms │
└─────────┴───────┴────────┴────────┴────────┴───────────┴─────────┴────────┘
┌───────────┬─────────┬─────────┬─────────┬────────┬─────────┬────────┬─────────┐
│ Stat      │ 1%      │ 2.5%    │ 50%     │ 97.5%  │ Avg     │ Stdev  │ Min     │
├───────────┼─────────┼─────────┼─────────┼────────┼─────────┼────────┼─────────┤
│ Req/Sec   │ 7703    │ 7703    │ 9543    │ 9687   │ 9190    │ 747.26 │ 7703    │
├───────────┼─────────┼─────────┼─────────┼────────┼─────────┼────────┼─────────┤
│ Bytes/Sec │ 2.07 MB │ 2.07 MB │ 2.56 MB │ 2.6 MB │ 2.46 MB │ 200 kB │ 2.06 MB │
└───────────┴─────────┴─────────┴─────────┴────────┴─────────┴────────┴─────────┘

Req/Bytes counts sampled once per second.
# of samples: 5

47k requests in 5.03s, 12.3 MB read

In my commit:

┌─────────┬────────┬────────┬────────┬────────┬──────────┬─────────┬────────┐
│ Stat    │ 2.5%   │ 50%    │ 97.5%  │ 99%    │ Avg      │ Stdev   │ Max    │
├─────────┼────────┼────────┼────────┼────────┼──────────┼─────────┼────────┤
│ Latency │ 140 ms │ 146 ms │ 165 ms │ 176 ms │ 147.6 ms │ 7.62 ms │ 210 ms │
└─────────┴────────┴────────┴────────┴────────┴──────────┴─────────┴────────┘
┌───────────┬─────────┬─────────┬─────────┬─────────┬─────────┬─────────┬─────────┐
│ Stat      │ 1%      │ 2.5%    │ 50%     │ 97.5%   │ Avg     │ Stdev   │ Min     │
├───────────┼─────────┼─────────┼─────────┼─────────┼─────────┼─────────┼─────────┤
│ Req/Sec   │ 6003    │ 6003    │ 6783    │ 6995    │ 6632.4  │ 363.27  │ 6000    │
├───────────┼─────────┼─────────┼─────────┼─────────┼─────────┼─────────┼─────────┤
│ Bytes/Sec │ 1.61 MB │ 1.61 MB │ 1.82 MB │ 1.87 MB │ 1.78 MB │ 97.5 kB │ 1.61 MB │
└───────────┴─────────┴─────────┴─────────┴─────────┴─────────┴─────────┴─────────┘

Req/Bytes counts sampled once per second.
# of samples: 5

34k requests in 5.03s, 8.89 MB read

i think the performance decreased, maybe using async await is not the best way or maybe i made some mistakes

[mcollina]

Hopefully performance will not be degraded when we are finished :/.

[mcollina]

how are the benchmarks?

[multivoltage]

at my commit:

┌─────────┬────────┬────────┬────────┬────────┬───────────┬─────────┬────────┐
│ Stat    │ 2.5%   │ 50%    │ 97.5%  │ 99%    │ Avg       │ Stdev   │ Max    │
├─────────┼────────┼────────┼────────┼────────┼───────────┼─────────┼────────┤
│ Latency │ 141 ms │ 153 ms │ 286 ms │ 324 ms │ 161.49 ms │ 32.4 ms │ 344 ms │
└─────────┴────────┴────────┴────────┴────────┴───────────┴─────────┴────────┘
┌───────────┬─────────┬─────────┬─────────┬─────────┬─────────┬────────┬─────────┐
│ Stat      │ 1%      │ 2.5%    │ 50%     │ 97.5%   │ Avg     │ Stdev  │ Min     │
├───────────┼─────────┼─────────┼─────────┼─────────┼─────────┼────────┼─────────┤
│ Req/Sec   │ 5003    │ 5003    │ 6463    │ 6995    │ 6090    │ 890.18 │ 5000    │
├───────────┼─────────┼─────────┼─────────┼─────────┼─────────┼────────┼─────────┤
│ Bytes/Sec │ 1.34 MB │ 1.34 MB │ 1.73 MB │ 1.87 MB │ 1.63 MB │ 239 kB │ 1.34 MB │
└───────────┴─────────┴─────────┴─────────┴─────────┴─────────┴────────┴─────────┘

Req/Bytes counts sampled once per second.
# of samples: 5

31k requests in 5.03s, 8.16 MB read

I start reading this issue

Using

const readFile = require("util").promisify(require("fs").readFile)

is really impressive and performance come back to the same than master commit. All my test are made with node 18. @mcollina what do you think about ?

[Uzlopak]

@mcollina @multivoltage

What about this PR?

[multivoltage]

@Uzlopak Hi. Since there was a lot of code new after some release I decided to create new branch from new release and did same commit from 0. So I have a ready branch but not pushed. That because after my last reply I did not find a method different from node utils promisify which keep same perfomance as original method readFile

[mweberxyz]

@multivoltage are you still in continuing with this work? I am interested in picking it up where you left off.

[multivoltage]

@mweberxyz nope in this moment. But in 2/3 days I will update my PR and maybe we can help each other if you want

[multivoltage]

@mweberxyz I updated my branch if you want to start from that. Two main points that are important for me:

• const { readFile } = require("node:fs").promises is VERY slow
• I ended with const readFile = require("node:util").promisify(require("node:fs").readFile); witch is similar do sync version, but a little bit slower
• in node js 21 test fails

[multivoltage]

thank @mweberxyz, I will close the issue :)

[Uzlopak]

Whats the point of using promisify on a sync function? You will anyway block the Event loop.

[multivoltage]

Only because async version of readFile seem to be slower than sync version. But nevermind since I closed PR :) in favor of new one by @mweberxyz