naive snark circuit and pow bits for compression wrapper #179
Conversation
| @@ -422,19 +423,19 @@ impl ZkSyncCompressionForWrapperCircuit { | |||
| use crate::boojum::cs::traits::circuit::CircuitBuilder; | |||
| match &self { | |||
| Self::CompressionMode1Circuit(..) => { | |||
| <CompressionMode1Circuit as CircuitBuilder<GoldilocksField>>::geometry() | |||
| <CompressionMode1ForWrapperCircuit as CircuitBuilder<GoldilocksField>>::geometry() | |||
There was a problem hiding this comment.
This may break existing functionality, I do not remember the path of how current scheduler proof is wrapper (most likely 1 FRI based proof to switch to BN254 hasher, and then PLONK proof), so such code path may be used and will break.
In general, would be nice to have comment on what's a difference between compression and compression for wrapper, especially at initial steps, as only one step should swap hasher
There was a problem hiding this comment.
In general, 'compression for wrapper' refers to using a BN-compatible hasher for both the transcript and the Merkle tree as it is the last step before EVM verification. Additionally, this is a circuit that verifies the proof of the previous ‘compression’ step. Therefore, it should return the geometry of 'CompressionMode1ForWrapperCircuit' for the constraint system, not 'CompressionMode1Circuit'.
What ❔
Compression wrapper mode uses 8 bits
Why ❔
PoW bits reduces number of FRI queries so that smaller snark verification circuit size.
Checklist
zk fmtandzk lint.