Don't use fixed nominator count for report_equivocation weight calculation

[andresilva]

This PR changes the report_equivocation weight calculation to take into account the runtime configured maximum number of nominators per validator, rather than using a constant value.

polkadot companion: paritytech/polkadot#7432

[davxy]

I struggle a bit with the idea that a consensus algorithm is so strictly bound to a staking concept such as nominators (that in practice is not mandatory to be used).

Isn't better to set this once and read from some external source?

(Just an idea) What about providing a WeightInfo implementation which has the ability to plug some external "extra cost" source (via a trait bound?). Staking pallet then implements this trait and can be used in any place to augment the weight with the cost related to the max nominators.

[andresilva]

@davxy I agree that currently the weight implementation is highly coupled with the existing staking implementation. We can probably abstract it in a way to decrease this coupling, although TBH I'm not sure how valuable that would be. Moving this to the staking pallet would make it depend on the internals of GRANDPA (basically reversing the coupling). And the WeightInfo is already pluggable in the pallet, so if you're using GRANDPA with another staking system you can already bring in your own weight implementation. Either way, feel free to create an issue to track this, I don't think such changes should be done in this PR.

[melekes]

👍

[andresilva]

bot merge