integrate DLEq into protocol

[athanorlabs]

integrate DLEq into protocol

[Lederstrumpf]

Of what's listed above, all but the integration into SendKeysMessage has already been completed during our hackathon in #4 - it's even serialized efficiently (64898 bytes), and also already includes the proof of knowledge mentioned by @kayabaNerve here: #60 (only improvement I can see so far would be stripping the ECDSA sig, which will save you 64 bytes + serialization overhead, if any).

Right now, the DLEQ proof is shared via a local file (dleq_proof.dat), which must of course be replaced, and the proof verification is handled via an external binary, which is not ideal. But if you integrate the serialized proof payload into the SendKeysMessage (which will then already work), I will make time to switch from the external binary to using FFI bindings for the DLEQ proof generation & verification. I believe what @alxs did here #27 will reduce effort too.

[Lederstrumpf]

another argument for going Rust FFI: @kayabaNerve's dleq lib has had a lot of thought put into abstracting different curves - I haven't properly reviewed it yet, but once I'm no longer swamped with other work, I'll integrate it into Farcaster too & adjust the FFI bindings here as well.

[kayabaNerve]

Lack of peer review is why it's still branched, at which point I'll publish it as a proper crate. If you write a new one in Go (briefly mentioned), I'd be happy to write tests running the two proofs against each other so we can define another implementation of a consistent protocol.

[noot]

@Lederstrumpf okay awesome, I can pick up on that work integrating it into the Go side. I'll also look into the FFI/CGO route and see what it'll entail exactly. I'll message you once the Go integration is done, then we can work on that. if that doesn't go well I'll probably try to port @kayabaNerve 's implementation to Go, briefly looking at the library, it seems it should be doable without too much pain.