Fork Wish List

[GalRogozinski]

Description

An ever-updated list of changes we want that need a fork.

List

In

• Network Topology #43 (performance & scalability)
• Aggregator Committee Duty #61 (performance & scalability)
• Remove Decided Messages from QBFT Consensus #63 (performance)
• Consider accounting Epochs for Round Robin proposer selection to introduce variance in operator selection across slots #65 (security)

Out

• Finality based event syncing  #67 (security) (because it isn't a forking change)
• If not QBFT leader, do not wait for start value - less important because currently no issues with proposals.
  Prerequisites:
  • Dynamic QBFT Timeout #37
  • Round change according to duty start time ssv-spec#352
• Handshake Improvements ssv-spec#532
• Optimize Msg Structures #58 - probably lots of work
• Do Proposal's pre-consensus as soon as possible in the start of the spec ssv-spec#373 - No issues with proposals
• Incorporate a scheme to replace RSA or fix PEM format #59
• Audit Suggestion 5: Unnecessary checks increase complexity ssv-spec#355

[GalRogozinski]

This list was continued in ssvlabs/ssv-spec#368

[moshe-blox]

@GalRogozinski requesting to add support for participating in consensus without data

given network topology reduces traffic and load, reducing it more is great but diminishing returns, so high effort optimizations like RSA->ecDSA can wait for next fork (IMO)

another vector to optimize is consensus time, and i think starting consensus before you got the block or attestation data would reduce it by a lot (and help avoid some cases of missed proposals)

wdyt?

[GalRogozinski]

@moshe-blox
you mean ssvlabs/ssv-spec#371? we can do it

[MatheusFranco99]

Fork: QBFT Identifier with fixed length -> msg encoding changes

[MatheusFranco99]

Remove repeated Signer field in spec.types.PartialSignatureMessage (already exists in spec.types.PartialSignatureMessages).