Degenerate Shielded Rewards Precision for IBC Tokens

[murisi]

Inflation is often a small fraction of the total amount of an asset that is locked. So if inflation were 5 and the locked amount were 100, then the reward would 5/100=0.05 NAM for every unit of an asset. Unfortunately decimals are not supported by the rewards circuit and the 0.05 would get truncated/floored to 0, so a workaround is applied where 0.05*precision NAM are rewarded for every precision T units of an asset. How is precision chosen for a given asset T? The shielded rewards algorithm currently uses a precision of 10^(max(denom,3)-3) under the assumption that 1 T is equivalent to a raw amount of 10^denom. Unfortunately, denom always equals 0 for IBC tokens (see

namada/crates/trans_token/src/storage.rs

Line 185 in 75059e7

Address::Internal(InternalAddress::IbcToken(_)) => {

) meaning that precision = 1 meaning that the reward fraction is floored to 0 just like the 0.05 example above. So probably no reward will be distributed for IBC tokens even if its parameters were set to be non-zero.