Allow querying for validator public key

[maurolacy]

Ideally, the pubkey would be returned as part of the Validator struct

https://github.com/CosmWasm/cosmwasm/blob/ebb378517fbf2c2093c873e5c99a2ea12ce640cf/packages/std/src/query/staking.rs#L131C10-L139

That means it'll be part of the StakingQuery::AllValidators and StakingQuery::Validator query responses.
It can be an optional field for backwards compatibility / non-breaking change.

Alternatively / non-ideally, add a new query type so that the pubkey can be queried.

We need this as part of Mesh Security. See by example osmosis-labs/mesh-security#88.

[webmaster128]

Do you need the consensus public key or the operator public key? In the Cosmos SDK query, only the consensus public key is easily available:
https://github.com/cosmos/cosmos-sdk/blob/v0.47.4/proto/cosmos/staking/v1beta1/staking.proto#L97-L98

What do you need the public key for?

[maurolacy]

The one that is used to sign blocks, that is the one we need. I think that would be the consensus one.

We need the validator public key in order to verify signatures on the Provider side, as part of cross-slashing.

[maurolacy]

See also: osmosis-labs/mesh-security#72.

[webmaster128]

I see. Yeah, this is google.protobuf.Any consensus_pubkey = 2 [(cosmos_proto.accepts_interface) = "cosmos.crypto.PubKey"];.

Adding this is source code breaking but we can do it as first thing of the CosmWasm 2.0 work and you can start using a development branch.

[webmaster128]

By the way, in https://twitter.com/Rarma_/status/1693977352823533666 you see a case where the operator address is not derived from a pubkey but is a CosmWasm contract address. This means you cannot rely on operator pubkeys/addresses to connect the same operator across chains. This matches the Cosmos SDK return data where you only get an operator address but not pubkey.

[maurolacy]

We do not need to connect operators across chains.

In fact, in the current design of mesh-security, we ended up not needing the public keys at all.

So, this can be closed. Or left open as a nice to have, if you want.

[webmaster128]

Thanks! I'll leave it open for now as it is a reasonable addition.

[chipshort]

Let's not increase the response of the list, but for querying a single validator it makes sense to add this to the response.
However, this requires a breaking change since both responses currently use the same type. Adding to the 3.0 milestone.

[webmaster128]

As an intermediate step we can also ensure to use separate Validator types in ValidatorResponse and AllValidatorsResponse. Then we can extend them later in a non-breaking way.