Remove usage of noble-secp256k1 tweak utils

[paulmillr]

I have removed them because I think their api is suboptimal. If you want to keep using them, copy-paste this:

https://github.com/paulmillr/noble-secp256k1/blob/37e66841eb155d21b4fe6658223140d5d7d5176e/test/index.ts#L467

[mmcshinsky-bitgo]

Thanks for the update! We're making the adjustments currently internally as well as a result of the change.

[paulmillr]

As a side note, your api which depended on those functions is also overengineered, e.g. return bigIntFromU8ABE(secp.utils.privateNegate(s)); could be just secp.utils.mod(-s, secp.CURVE.n). Every type conversion has potential error somewhere, which means something could go wrong.

  scalarSub(x: bigint, y: bigint): bigint {
    const negatedY = secp.utils.privateNegate(y);
    return bigIntFromU8ABE(secp.utils.privateAdd(x, negatedY));
  }
  // could just be
    scalarSub(x: bigint, y: bigint): bigint {
    return secp.utils.mod(x - y, secp.CURVE.n)
  }

[mmcshinsky-bitgo]

That's great to know, thanks!

@brandonblack You may be interested in this.

[brandonblack]

Thanks @paulmillr.

FWIW, removing these in a minor version caused us quite some pain when builds started spontaneously failing due to semver ^.

[paulmillr]

@brandonblack sorry for this, I intend to follow semver, however, I did not even add these methods to docs, and considered them experimental. What should have been done from my end: naming them with _underscore. That should have been better communicated from my end.

[brandonblack]

Thanks. We'll update to 1.7.0 and expand our wrapper to include these functions.

[mmcshinsky-bitgo]

@brandonblack What is the status of this issue? Is this issue now resolved?

[brandonblack]

@mmcshinsky-bitgo no. will probably work on it next sprint - this is only an issue when we upgrade noble-secp256k1 to 1.7+