← All calls
ARBORIST#134

ZSA catches up to NU6.3, Epoch team joins

ZSA finished catching up to NU6.3, and Shielded Labs launched Epoch, a post-quantum research team.

Agenda summary 8

  1. Zebra update

    Zcash Foundation · Marek

    • The team plans to ship several optimizations for mining pools before NU7 activates. The NU7 activation height is expected to be set next Tuesday.
  2. Core libraries and Zallet

    Zodl · Pacu

    • Over the past two weeks the team aligned the NU7 pre-release of the librustzcash crates and zcashd with the new network upgrade, and did a lot of ZIP reviews.
    • It released pasta_curves 0.6.1. Assembly field operations for AArch64 and x86-64 greatly improved performance.
    • Because the Zingo side is making many changes, the Zaino backend was removed from Zallet for now and will be reattached later.
    • A Zallet release for the NU7 testnet is in preparation, and a mainnet release will follow once the activation block height is set. Documentation was also expanded.
    • The new lightwalletd version added a metric that reports stale tips, and the Docker image shrank from about 2GB to under 200MB, roughly a 90% reduction, according to the explanation. The PCZT crate for the Ledger feature was also polished.
    • The team said it had scheduled the grant payments for October 16.
  3. Zcash Shielded Assets

    QEDIT · Vivek

    • The team finished catching up with the network upgrades that led to the recent security incidents, so ZSA is now aligned through NU6.3.
    • The work was posted to the Orchard repository in two parts. Upstream PR 506 contains only the circuit changes, in 37 files, to make it easy to review, while the builder and the rest are in an internal QEDIT PR.
    • librustzcash and Zebra are also aligned to NU6.3.
    • The handover of the work from the QEDIT repository to a dedicated branch in the Zcash repository is being discussed with the core team. The team said a grant adjustment may be needed because of the added NU6.2 and NU6.3 work.
    • At the end, Daira said they would give access for merging the branch.
  4. Crosslink and the Trailing Finality Layer

    Shielded Labs · Philip

    • The first FeatureNet restart finished without problems on October 1. The BFT bootstrap succeeded, and BFT is now active and finalizing.
    • The 10% finalizer fee is paid only to finalizers on the active roster. In this FeatureNet only the top 12 are active, and that number is for testing. A staking day is a real 24 hours, which is about 3000 blocks at 25-second blocks.
    • The second payout of the incentivized testnet distributed 25 ZEC. About 20 ZEC went to mining and staking participants, and 5 ZEC was paid at discretion to contributors of tooling, testing and community support.
    • Node software version 15 was released along with a small network upgrade. It fixed a BFT exception that made finalizers stall when there were multiple proposals in the same round.
    • The finalizer fee issuance model was also corrected to work as intended. The bond amount is enforced by consensus to be a power of 10, and it activates in about 6 hours, near block 55,000.
  5. Dynamic fees

    Shielded Labs · Mark

    • A ZIP draft of the fee estimation mechanism is expected around next week. The team plans to recommend to wallets helping with the NU7 transition that they implement the getStandardFees RPC and a priority lane that pays a higher fee.
    • Daira said they had heard that Zakura uses a minimum fee different from the ZIP and asked about the status. Mark said they understood Zakura was proceeding with 400 zat and that Arya had, as far as they knew, opened a PR to amend the ZIP.
    • Daira pointed out that the fee reported by a light wallet server depends on the node behind it, so paying a lower fee becomes an interoperability problem.
    • Marek said the Zebra side has not decided yet, and Daira regretted that no Zakura representative was present.
  6. Formal verification and post-quantum research

    Zodl · Daira

    • The speaker is attending the Cambridge formal verification workshop. In pasta_curves, the AArch64 and x86-64 Montgomery multiplication and the AArch64 and portable inversion were formally verified. The assembly came from Zakura.
    • The speaker presented keywords and macros for checking that functions are efficient when expressing security reductions in Lean.
    • The Ironwood formalization covers the Orchard and Ironwood action circuits, Halo2 knowledge soundness, and balance integrity, and the next goal is spendability.
    • Shielded Labs is also doing Quint modeling for its zero indexer tool and plans to publish a blog post soon.
    • The speaker thinks Akita, a lattice-based post-quantum polynomial commitment, could be used in place of Ragu's PCS. The estimated proof size is about 85k, roughly 12 times that of the discrete-log approach and smaller than the roughly 200KB or more of hash- and code-based schemes.
    • Research remains to be done, since part of the security proof is missing and a small design change is needed to fit Ragu. The estimation tool was posted in the chat.
  7. Epoch

    Shielded Labs · Ulrich

    • Ulrich joins Shielded Labs on October 1 to lead the Epoch team. Two former Aztec members (Suez Spaga and Luke Edwards) are joining, making a team of three.
    • Epoch's goal is to define the proof and signature schemes and the overall design to prepare for post-quantum threats. It will also address the throughput problem caused by proofs becoming at least 12 times larger.
    • The team plans to collaborate with ecosystem players such as Zodl.
  8. Announcements and discussion

    Marek

    • Strad found a contradiction between ZIP 218 and ZIP 237. If NU7 ships as is, rounding would remove 0.67 ZEC from the total issuance.
    • Either leave it as is and fix it in a later upgrade, or fix it now, which would require rolling back the testnet. Daira said they would amend ZIP 237 so that its claim is corrected by ZIP 218.
    • The next Arborist call is on November 12. The October 29 call, which falls during ZCon, was canceled.

This summary is based on the original meeting notes; check them for exact details.