PQ Interop #029
Transcript
Ask about this call
Answered from this call's summary and transcript only.
AI-generated — verify against the recording.
Call summary
Targets
- •End of week - DevNet-4 spec completion target — 00:52:11
- •This week - Gene team wrapping up DevNet-2 work — 01:00:22
Decisions
- •DevNet-5 will use separate keys for block proposing and attesting (not ZK approach) — 01:28:38
- •Short-term: use even/odd XMSS indices; long-term: dual key system with independent lifetimes — 01:30:51
- •Do not assume zero-knowledge properties for consensus layer design — 01:37:13
Highlights
- Infrastructure:
- ·Subnet count hardcoded as 1; needs config-based approach for multi-subnet testing — 01:07:08
- ·Plan to scale validators beyond minimal counts for realistic testing — 01:10:53
- ·DevNet recovered from 500-slot stall to 2,500 slots via checkpoint sync — 01:39:11
- Organizational:
- ·Kamil stepping down as Quadrivium CEO; joining Ethereum Foundation — 00:55:25
- ·Fort Mode event April (analogous to Beam Day); focus on consensus research — 01:41:38
- Devnet 3 Status:
- ·5-client interop achieved: Zeam, Ream, Lantern, Ethlambda, leanSpec — 00:51:38
- ·DevNet-3 merged to main in Zeam; ready for long-running testnet — 00:52:02
- ·Checkpoint sync issues resolved across clients; further stability testing needed — 00:55:59
- Devnet 5 Design:
- ·DevNet-5 proposal: sign both block and proposer attestation using recursive aggregation — 01:13:03
- ·Security concern: current proofs are succinct but not zero-knowledge — 01:15:18
- ·Alternative: aggregate signatures with different messages per party using Snarks — 01:22:00
- ·Consensus: use separate keys for proposer/attester; simplest secure approach — 01:28:29
- Devnet 4 Progress:
- ·DevNet-4 spec PR up for review; recursive aggregation via leanMultisig — 00:52:38
- ·Blocked on LeanSig ZK-friendly encoding implementation for recursive aggregation — 00:53:23
- ·ZK-friendly encoding not yet implemented; timeline uncertain due to workload — 01:04:17
Action Items
- •All client teams — Review DevNet-4 spec PR and provide feedback — 00:52:29
- •Anshal (Zeam) — Move subnet count from hardcoded to chain config — 01:07:38
- •Spec team — Spec DevNet-5 with separate proposer/attester keys approach — 01:33:25
Meeting chat44
- read.ai meeting notes
Kai added read.ai meeting notes to the meeting. Read provides AI generated meeting summaries to make meetings more effective and efficient. View our Privacy Policy at https://www.read.ai/pp Type "read stop" to disable, or "opt out" to delete meeting data.
- Bubbles Notetaker
Hi 👋 I'm recording this meeting and taking notes for Tamaghna Choudhuri. After the call, you'll get an AI summary and action items by email. Try Bubbles for free🔥 — Turn on automatic meeting note-taking: https://bubbl.es/notetaker
- this meeting ➡️
https://bubbl.es/remove-notetaker?t=PmD5nh62scAaSk2rew9pbxQnUf4HfkVjN
- Gajinder Singh
gm gm
- Will Corcoran
morning all. we will start in 2min
- Will Corcoran
Devnet-4 pr: https://github.com/leanEthereum/leanSpec/pull/426
- Will Corcoran
- Gajinder Singh
Anshal will be able to update on that
- Will Corcoran
Congrats, Kamil
- Mercy Boma Naps-Nkari
Reacted to "Congrats, Kamil" with ❤️
- Mercy Boma Naps-Nkari
Reacted to "Congrats, Kamil" with 🔥
- Kamil Salakhiev
Reacted to "Congrats, Kamil" with ❤️
- Gajinder Singh
Reacted to "Congrats, Kamil" with ❤️
- Parthasarathy Ramanujam
The checkpoint sync issue seem to be consistent across all clients. It appears to be working now on zeam. But, we will need further tests to ensure this fix is stable. @Katya should we create a common issue to monitor this?
- Parthasarathy Ramanujam
Reacted to "Congrats, Kamil" with ❤️
- Will Corcoran
Replying to "Also: https://github..." Is there anything blocking the devnet-4 spec / pm PRs?
- Gajinder Singh
Replying to "Also: https://github..." @Anshal
- Will Corcoran
Reacted to "Anshal will be able ..." with 🙏
- Gajinder Singh
also for spec, readiness for leanmultisig for recusiveness
- Shariq
Amazing 👍
- Parthasarathy Ramanujam
Are we going to retain a single subnet for devnet4 as well?
- Gajinder Singh
Replying to "Are we going to reta..." we will have multiple subnets even for devnet3 testing
- Parthasarathy Ramanujam
Replying to "Are we going to reta..." Devnet3 spec hardcodes the subnet as 1
- Will Corcoran
Bag it. Tag it
- unnawut
I think subnet_id is currently this, it’s dynamic but num_committees need to be bumped up https://github.com/leanEthereum/leanSpec/blob/44f896af5f43f71cdafaa993b61b4f4aa99681c0/src/lean_spec/subspecs/containers/validator.py#L31-L40
- Mercy Boma Naps-Nkari
It should be code lab with utah
- Gajinder Singh
Replying to "I think subnet_id is..." cool
- Mega | Lambda
Why do we want the proposer to both sign the block and the proposer attestation?
- Mihir Faujdar
If the proposer signs over the block, could that be treated as a proposer attestation?
- Gajinder Singh
Replying to "If the proposer sign..." hw can next proposer aggregate it?
- Kamil Salakhiev
curious how this works with bls today, will check later
- Will Corcoran
Replying to "If the proposer sign..." So, the goal is to have a single proof/signature that can be used to both: a) verify the block b) and to aggregate proposer attestations cleanly with other attestations in the network ?
- unnawut
Replying to "curious how this wor..." BLS can sign multiple messages and produce separate signatures at any time, not subject to one time sig constraint
- Parthasarathy Ramanujam
Replying to "curious how this wor..." The issue is due to OTS
- unnawut
Reacted to "The issue is due to ..." with 👍
- Mihir Faujdar
Reacted to "BLS can sign multipl..." with 👍
- Mihir Faujdar
Reacted to "The issue is due to ..." with 👍
- Kamil Salakhiev
Replying to "curious how this wor..." yes, but does proposer sign the same message that attesters sign today?
- Kamil Salakhiev
Replying to "curious how this wor..." if so, we should try to do the same
- Parthasarathy Ramanujam
These proposer keys are distributed via p2p?
- Mihir Faujdar
Reacted to "So, the goal is to h..." with 👍
- unnawut
Btw. Separate keys = good first step towards proposer-attester separation too? :S
- Gajinder Singh
Reacted to "Btw. Separate keys =..." with 😀
- Gajinder Singh
Replying to "Btw. Separate keys =..." good food for thought!
Key decisions
DevNet-5 will use separate keys for block proposing and attesting (not ZK approach)
Alternative to recursive aggregation with ZK-friendly encodingShort-term: use even/odd XMSS indices; long-term: dual key system with independent lifetimes
Key management strategy for XMSSDo not assume zero-knowledge properties for consensus layer design
Consensus layer design considerations