new Ethers.Wallet giving me a 132 character public key?

[nolanjannotta]

Hey guys I need help understanding something. When I use wallet = new Ethers.Wallet() and pass in a private key, and then wallet.publicKey, I am given a public key with 132 characters ( INCLUDING the 0x, so 130 without) Everywhere I read says a public key is 128 characters without the 0x. Eth.Build gives me a public key of 128 characters as a reference. any idea what I'm missing?

Thanks

[ricmoo]

A public key (uncompressed) is 65 bytes, which is 130 nibbles. The first byte for an uncompressed key is always 04, which indicates the key is uncompressed. The first by of a compressed key is always either 02 or 03, depending on whether the y coordinate is “positive” or “negative” (the first bit is 0 or 1 respectively).

If you are using a library that expects just the raw x and y, you can trim off the first byte (the 04), but most libraries should be able to understand that initial byte and length indicates the following key is uncompressed and trim it off for you.

Make sense?

[ricmoo]

(The standard is a public key should be 65 bytes for an uncompressed key and 33 bytes for a compressed key; the Ethereum algorithm for computing an address however, uses a non-standard representation internally, which is likely what Eth.Build is returning)

[nolanjannotta]

That does make sense! thanks for replying. so, uncompressed and compressed public keys can be used interchangeably?

[ricmoo]

For ethers and most libraries, yes; they are trivial to convert between if you have a secp256k1 library at your disposal, so most libraries will automatically convert the values internally. :)

If you have a raw 64 byte (128 nibble) public key, you can convert it to a standard form using ethers.utils.concat( [ "0x04", rawKey ] ).

If you pass a key into any function or method in ethers that supports public keys, either (compressed or otherwise) will work.

Quick background how point compression works: A public key is just a point (x, y) on the elliptic curve. For any given x on the curve there are at most 2 solutions, you can think of it as y and -y which satisfy the equation. So, from x, we can compute both y and -y, so we have 2 solutions (x, y) and (x, -y). Rather than include x and y, we store only the x and include 1 extra bit to indicate whether we want y or -y (both of which we can compute from x), which is why we need that extra byte, for that single bit. A prefix of 0x02 means compute y as the last 32 bytes of the public key, and 0x03 means compute -y.

[nolanjannotta]

wow, I didn't know that, thats interesting. thanks!