text: mention that index is required for signing

[jonasnick]

Fixes #38

[real-or-random]

So you're suggesting not to change the DKGOutput to include the index, right? Can you elaborate on why not?

[jonasnick]

So you're suggesting not to change the DKGOutput to include the index, right? Can you elaborate on why not?

Correct. The reason to include the index in the DKG output would be that this would make the DKG output contain sufficient information for signing. However, if the signer wants to initiate a signing session, this DKG output would not be sufficient because the signer needs to know also the indices of the participating signers. Hence, even for that DKG output the signer (and coordinator) would have to have a way to get the index of a particular hostpubkey from the list (if they want to initiate a signing session).

Also, finding the index of a hostpubkey in the list is trivial (and it's in some sense already input of the DKG).

[jesseposner]

Also, finding the index of a hostpubkey in the list is trivial (and it's in some sense already input of the DKG).

Is this because the ordering of the hostpubkeys in the list determines the index assignment?

[jonasnick]

Yes

[real-or-random]

Also, finding the index of a hostpubkey in the list is trivial (and it's in some sense already input of the DKG).

Okay, that makes sense. Also with the signing BIP in mind, which uses indices instead of arbitrary identifiers.

If the signing BIP were to use arbitrary identifiers instead, we'd probably want to output our choice of identifiers. But if everything uses indices, then it's okay to leave this implicit. (The only caveat is that signing takes as input 32-byte arrays that will be interpreted as integer indices. I think it's a bit more natural to use integers directly, but perhaps this could be changed in the signing BIP)

[real-or-random]

(The only caveat is that signing takes as input 32-byte arrays that will be interpreted as integer indices. I think it's a bit more natural to use integers directly, but perhaps this could be changed in the signing BIP)

Just noticed that this is tracked in siv2r/bip-frost-signing#9 already.