web3.js: use Uint8Array instead of Buffer for Transaction.addSignature

[macalinao]

Problem

Buffer is used where Uint8Array is sufficient, making Browserify required for simple libraries.

See: https://twitter.com/armaniferrante/status/1569871757687029761

[codecov]

Codecov Report

Merging #27861 (b7759d4) into master (e6687b8) will increase coverage by 0.0%.
The diff coverage is 100.0%.

@@           Coverage Diff           @@
##           master   #27861   +/-   ##
=======================================
  Coverage    77.3%    77.4%           
=======================================
  Files          55       55           
  Lines        2889     2910   +21     
  Branches      402      404    +2     
=======================================
+ Hits         2234     2253   +19     
- Misses        514      515    +1     
- Partials      141      142    +1     

[jordaaash]

This looks fine to me, but this will fail commitlint, along with all the other PRs. I have a few asks --

1. Can you prefix the commit messages so CI will run? fix: should be fine for nonbreaking stuff.
2. Can you mark PRs that are breaking? For example, a change to arguments from (bytes: Buffer) to (bytes: Uint8Array) is nonbreaking, but changing the types/returns of public properties/methods is, so we have to figure out what to do about that.
3. Can you logically group these PRs somewhat? Reviewing like 20 PRs, many of which are related, is tougher to do in isolation.

[macalinao]

This looks fine to me, but this will fail commitlint, along with all the other PRs. I have a few asks --

1. Can you prefix the commit messages so CI will run? fix: should be fine for nonbreaking stuff.

Sure

2. Can you mark PRs that are breaking? For example, a change to arguments from (bytes: Buffer) to (bytes: Uint8Array) is nonbreaking, but changing the types/returns of public properties/methods is, so we have to figure out what to do about that.

Sure, will use conventional commits here

3. Can you logically group these PRs somewhat? Reviewing like 20 PRs, many of which are related, is tougher to do in isolation.

What grouping would give the PRs the highest probability of being accepted?

[jordaaash]

What grouping would give the PRs the highest probability of being accepted?

Well... it's hard to say without looking at all of them! I'd say maybe start by grouping the ones having to do with Buffer -> Uint8Array together, especially any that are breaking. There are some non-obvious ways a change like this can be breaking: #27700

In principle, I think we're all for getting away from Buffer, but I think we have to see if ultimately it's going to mean shipping web3.js v2.0 just for this.

[steveluscher]

Begone, typed array!

[steveluscher]

We can go straight into bs58.encode() as an array of numbers!

Suggested change

	signatures.push(bs58.encode(Uint8Array.from(signature)));
	signatures.push(bs58.encode(signature));