Compare BIP-44 and PR-15

[DavidM-D]

I've proposed a way of doing key derivations in PR-15. Serhii quite reasonably pointed out that I'm largely reinventing BIP-44.

We should always err on the side of not making new serialization formats. Can we achieve our UX goals with BIP-44, if not are there commonly supported extentions that do support that. If neither of these are possible, can we use parts of BIP-44 like SLIP-44 registered coin types?

[DavidM-D]

One thing to consider as well is that we can do BIP-44 compatible key derivation strings, but I don't believe we can do BIP-39 compatible derivations, so the public keys won't be developed in a predictable way for now, but there exists a PoC implementation that does this.

[DavidM-D]

Added some context to the choice here.

https://github.com/near/near-fastauth-wallet/blob/dmd/chain_sig_docs/docs/chain_signature_api.org#borsh

[diogoortega]

The main issue I see with BIP-44 is that we wouldn't be able to include domains, which we need to separate accounts used by different apps/domains. I think a more flexible key derivation structure could benefit the user experience overall, even for future use cases, so I'd be in favor of keeping it as described here.