Word64 is the type that both of the EpochNo and EpochSize are based on. It especially prone to integer overflow when doing arithmetic. For this reason it would be safer to remove Num instance alltogether for those types from the main library
This will be even less of an inconvenience for EpochNo, because in ledger we have introduced EpochInterval type and addEpochInterval :: EpochNo -> EpochInterval -> EpochNo function that make the Num instance less useful.
The only downside to removing the Num instance is that construction from literals requires usage of a constructor, but that is only a tiny burden for testing.
Therefore, instead of completely removing the Num instances, I suggest we move them as standalone orphans into the new testlib, which would allow us to have the cake and eat it too.
As part of this task we should also move EpochInterval and addEpochInterval from cardano-ledger into the cardano-slotting
Word64is the type that both of theEpochNoandEpochSizeare based on. It especially prone to integer overflow when doing arithmetic. For this reason it would be safer to removeNuminstance alltogether for those types from the main libraryThis will be even less of an inconvenience for
EpochNo, because in ledger we have introducedEpochIntervaltype andaddEpochInterval :: EpochNo -> EpochInterval -> EpochNofunction that make theNuminstance less useful.The only downside to removing the
Numinstance is that construction from literals requires usage of a constructor, but that is only a tiny burden for testing.Therefore, instead of completely removing the
Numinstances, I suggest we move them as standalone orphans into the newtestlib, which would allow us to have the cake and eat it too.As part of this task we should also move
EpochIntervalandaddEpochIntervalfromcardano-ledgerinto thecardano-slotting