The current de-dup mechanism is rather limited. Right now, it will only detect duplicates that exist in consecutive rounds. However, given the current design for interfacing with consensus, this might not catch all/most duplicates. In another issue, it is suggested we use a caching mechansim to fill gaps in nonce sequences. We could use a very similar mechanism to implement a more general de-dup mechanism.
Related to this, there seems to be a minor flaw in the spec, in that it says that "Members must keep the following state between rounds: ..." That state is insufficient to implement the stated de-dup strategy
The current de-dup mechanism is rather limited. Right now, it will only detect duplicates that exist in consecutive rounds. However, given the current design for interfacing with consensus, this might not catch all/most duplicates. In another issue, it is suggested we use a caching mechansim to fill gaps in nonce sequences. We could use a very similar mechanism to implement a more general de-dup mechanism.
Related to this, there seems to be a minor flaw in the spec, in that it says that "Members must keep the following state between rounds: ..." That state is insufficient to implement the stated de-dup strategy