CDDL #420
CDDL #420
Conversation
| , endorser_block_hash : hash32 ; Hash of the endorsed block (EB) | ||
| , persistent_voters : [* persistent_voter_id] ; Set of persistent voter IDs | ||
| , nonpersistent_voters : {* pool_id => bls_signature} ; Non-persistent voters with eligibility proofs | ||
| , ? aggregate_elig_sig : bls_signature ; Optional aggregate eligibility signature |
There was a problem hiding this comment.
I'm wondering if "optional" is misleading: It is required if there are non-persistent voters. However, in some stake-distribution scenarios there might not be any non-persistent voters, in which case it is absent.
There was a problem hiding this comment.
How about changing the comment to clarify its conditional existence based on the voter composition?
, ? aggregate_elig_sig : bls_signature ; Aggregate eligibility signature (present when non-persistent voters exist)
| bls_signature = bytes .size 48 ; BLS12-381 G1 signature (compressed) | ||
| bls_vkey = bytes .size 96 ; BLS12-381 G2 public key (compressed) |
There was a problem hiding this comment.
My memory is that other parts of the Ledger use store BLS points uncompressed. I never understood why and I think that Leios should store them compressed. Decompressing them does take a bit of computation, however, because the y coordinate needs to be looked up on the curve. (Presumably a similar computation is needed for an uncompressed point, in order to ensure that it truly is on the curve.)
There was a problem hiding this comment.
According to CIP-0381 it seems ok to use compressed and is recommended.
"even though uncompression is a lot more expensive per point than deserialisation, the size savings due to compression actually outweigh the speed gains due to serialisation because bytes per script are a lot more expensive than ExUnits per script in real terms."
|
For the IB and EB CDDL, we'll need to decide whether a node is allowed to produce multiple IBs in the same slot or multiple EBs in the same pipeline.
Personally, I think the probability of being allowed multiple IBs or EBs is small enough (even for the biggest stakepools) that we should limit it to one. This simplifies the data structures, makes it easier to detect equivocation, etc. |
I agree. It would also simplify the network protocol presumably (by disallowing multiple announcements per slot). |
Uh oh!
There was an error while loading. Please reload this page.