feat: update to new libp2p interfaces

[achingbrain]

Updates this module to use the @libp2p/* set of modules as part of the typescript port.

• Renames imports to add .js suffix (tsc does not add this during compilation to ESM)
• Renames bundled @types/*.d.ts to @types/*.ts because tsc ignores .d.ts which masks errors
• Updates project config to output ESM instead of CJS code
• Updates all deps to lastest versions including ones that have gone ESM-only
• Re-generates protobuf code with the ES6 target
• Fixes all linting errors
• Swaps bl for Uint8ArrayList

N.b. pbjs generates invalid imports for bundler-less ESM - the .js suffix has to be added to the import manually so it's best not rebuild the protobuf generated code unless you know something has changed.

BREAKING CHANGE: This module now outputs ESM code only

[achingbrain]

Running the benchmarks the performance is similar, slightly slower:

New:

$ node benchmark.js 
Initializing handshake benchmark
Init complete, running benchmark
handshake x 52.53 ops/sec ±16.17% (72 runs sampled)

Old:

$ node benchmark.js 
Initializing handshake benchmark
Init complete, running benchmark
handshake x 58.42 ops/sec ±15.39% (77 runs sampled)

[codecov-commenter]

Codecov Report

Merging #134 (93b1c16) into master (790b039) will decrease coverage by 4.70%.
The diff coverage is 76.81%.

❗ Current head 93b1c16 differs from pull request most recent head e78bfd9. Consider uploading reports for the commit e78bfd9 to get more accurate results

@@            Coverage Diff             @@
##           master     #134      +/-   ##
==========================================
- Coverage   85.91%   81.21%   -4.71%     
==========================================
  Files          16       26      +10     
  Lines        1839     2864    +1025     
  Branches      235      279      +44     
==========================================
+ Hits         1580     2326     +746     
- Misses        259      538     +279     

Impacted Files	Coverage Δ
test/noise.spec.ts	52.34% <48.91%> (ø)
test/handshakes/xx.spec.ts	90.62% <55.55%> (ø)
src/handshakes/abstract-handshake.ts	89.65% <57.14%> (-2.43%)	⬇️
test/ik-handshake.spec.ts	55.95% <68.42%> (ø)
test/handshakes/ik.spec.ts	95.83% <71.42%> (ø)
src/proto/payload.js	78.64% <78.43%> (-1.08%)	⬇️
src/utils.ts	90.66% <80.00%> (-7.74%)	⬇️
test/utils.ts	90.00% <80.00%> (ø)
src/noise.ts	76.22% <83.33%> (-0.29%)	⬇️
test/xx-fallback-handshake.spec.ts	93.58% <85.71%> (ø)
... and 38 more

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 790b039...e78bfd9. Read the comment docs.

[dapplion]

Thank you so much! before merging this Lodestar must switch to ESM too, so we'll wait for that to happen

[achingbrain]

I'm not sure it's necessary to wait, this can (and should) be published as a major, Lodestar can use the previous version until it's ready to upgrade?

[dapplion]

Looks good to me! @achingbrain yeah you are right we can merge this and keep using the current version is Lodestar. And backport fixes if necessary.

[mpetrunic]

This is so weird. Is there a reason why .d.ts is not ok anymore like it should not be compiled to js

[achingbrain]

From the way I understand it, .d.ts files are meant to be machine generated so they aren't parsed and validated like regular .ts files.

We've certainly shipped a lot of broken types in the past due to hand-rolled .d.ts files - things like import paths being wrong etc don't trigger errors so it's a lot safer this way.

[achingbrain]

Will this get a release soon? It's the last piece of the puzzle before I can open the ts PR for libp2p itself.

[mpetrunic]

will do now

[mpetrunic]

Will this get a release soon? It's the last piece of the puzzle before I can open the ts PR for libp2p itself.

https://github.com/ChainSafe/js-libp2p-noise/releases/tag/v6.0.0