← All calls
ARBORIST#127

NU 6.3 testnet targeted for July 1

Activation of the NU 6.3 testnet, which includes Ironwood, was targeted for July 1.

Agenda summary 13

  1. Zebra update

    Zcash Foundation · Arya

    • The release process was further automated with a consolidated release workflow, and the team started attaching signed binaries as release artifacts.
    • The reorg window of the non-finalized state was increased from 100 blocks to 1000 blocks, so the most recent 1000 blocks are kept in memory and can be rolled back.
    • A find fork point function contributed by Kris finds the height and hash of a fork's common root.
    • Resumable non-finalized block streaming was added to the index RPC server, so Zaino and Zallet sync better with the trusted chain sync module.
    • Work started on the NU 6.3 network upgrade, including the Ironwood value balance pool.
    • The team is also adding zcashd's block notify configuration option, which runs a command when the chain tip changes.
  2. ZODL update: Ironwood

    ZODL · Pacu, Daira Emma

    • For Ironwood's v6 transaction format and signing implementation, the team reviewed and merged Valar Group's implementation.
    • It also reviewed the Ironwood-related librustzcash PR sent by Shambo and Valar Group's implementation of Keystone multi-transaction signing (for moving funds).
    • The day before, the drafts of the NU 6.3 deployment ZIP and the v6 transaction format ZIP were merged and can be seen on the ZIP site.
    • The target for NU 6.3 activation on testnet is July 1, but the height is left TBD to be set at the last moment, since miners will test with powerful equipment.
    • Daira Emma said the NU 6.3 circuit changes were merged into the Orchard crate, and what remains is two ZIPs covering migration and wallet considerations and the ZIP 317 update on fees.
    • The consensus changes of the emergency upgrades just before NU 6.2 and 6.3 are being documented in the specification.
    • Documentation and specifications on the user experience and the future Ironwood migration are also being written.
  3. ZODL: zcashd, core libraries and Zallet update

    ZODL · Pacu

    • The zcash_primitives 0.25 dependency upgrade was completed across the stack.
    • Zallet alpha 4 is complete and going through the release process, and may be public that day.
    • Because it includes zcashd data file migration in addition to the new wallet path, the team asked people to back up their files, try the migration on testnet or mainnet and report any balance differences.
    • Zallet will soon offer an option to choose either Zaino or Zebra directly as the backend.
    • The problem of mutually exclusive features colliding when building with all features will be resolved then.
  4. Zaino update

    Zingo Labs · Za

    • The team released Zaino 0.4.1, which includes the NU 6.2 upgrade.
    • It resolved a bug that was tangled up with the experimental gettxoutsetinfo feature and was causing sync problems.
    • It is preparing 0.5.0, which includes an optional database flag that lets it be used without waiting for index construction, and this also addresses the gettxoutsetinfo sync problem.
    • Zingo Lib, which depended on Orchard directly and caused dependency conflicts, was removed entirely as a test dependency and turned into a dev tool.
    • The default zcashd tests were turned off and made opt-in with a zcashd flag.
    • The release criteria are being changed to center on live tests of Zallet's real use cases, and a Zallet CLI will be added to the harness.
    • An upgrade to the latest Zebra, which uses the 1000-block reorg depth, is also underway.
  5. zcashd deprecation update

    Electric Coin Company, Zcash Foundation, Zingo Labs · Pacu, Pili

    • The PR for the option to choose between Zaino and Zebra as Zallet's backend will be merged soon.
    • It is not certain whether it will be included in alpha 4.
    • Pili highlighted Zebra's block notify configuration option.
    • It is a feature the team did not know people were using, but it is now implemented.
  6. ZSA update

    Qedit · Vivek

    • Qedit's ZSA branch of Orchard has caught up with all of the NU 6.2 changes and is at the final review stage.
    • Since ZSA is a later upgrade and does not need to work on a bad circuit, the handling of circuit versions is being cleaned up.
    • PR 10628 was opened in Zebra, and the team asked ZF to review the high-level approach to see whether a major change of direction is needed.
    • Ironwood and security work have higher priority, so the team will respond when feedback comes.
    • The ZIPs are watching the effect of Ironwood's new transaction format, the extensible transaction, and quantum recoverability on ZSA and swaps.
    • librustzcash and the transaction tool are being kept ready to respond to the changes.
  7. Network Sustainability Mechanism (NSM) update

    Shielded Labs · Judah

    • Two PRs are open: the ZIP 234 PR and a ZIP 235 bug fix PR.
    • The speaker said time may be short with Ironwood in progress, but a review would be appreciated.
  8. Crosslink update (Trailing Finality Layer)

    Shielded Labs, ZODL · Nate, Daira Emma

    • The feature net remains in its initial generation with finality stalled, and some users are in a permanent fork they cannot resolve because of the 100-block rollback limit.
    • Instead of resetting, the team is analyzing these unfortunate cases and improving the tooling so users can identify the cause.
    • Mainnet rewards for the first season are split into two kinds.
    • One is a non-discretionary reward proportional to testnet issuance, and the other is a discretionary reward from Shielded Labs to community contributors who built infrastructure and tooling and to active participants.
    • On the software side, the team is testing a user-driven slashing tool.
    • It is an exploratory tool that helps enough users reach agreement and organize a fork that slashes a particular finalizer's stake.
    • Improvements that are not backward compatible are queued up and will be deployed as the second generation when the network is rebooted.
    • When Daira Emma asked why user-driven rather than automatic slashing was chosen, Nate answered that automatic rules have false-positive and false-negative problems, and that availability slashing can be abused for DoS, so user-driven was chosen for simplicity.
    • Daira Emma said they had also heard the standard arguments for automatic slashing, and deferred the deep discussion to the R&D meeting and the Zcash Summit.
    • For the ZIP draft, the basic design was written in which issuance is distributed through the PoS subsystem to bond holders and as finalizer fees.
    • The third workshop is tentatively planned for July 23.
  9. Dynamic fees update

    Shielded Labs · Mark

    • Cipher Wallet (the wallet from the makers of Cipher Pay and Cipher Scan) will implement the priority and acceleration semantics feature within a week, and the browser extension wallet Noir Wallet will implement it too.
    • Priority transaction activity on mainnet is expected to increase.
    • A PR was created to add a getstandardfee RPC to Zebra.
    • For now it returns only the ZIP 317 marginal fee of 5000 zat, and the intent is for users to use the same endpoint later when a fee estimator is released.
    • Alfredo and Gustavo reviewed it.
    • The team also plans to build a public visualization site showing the priority lane.
  10. Grant presentation: formal verification of zebra-chain

    Runtime Verification · Natalie, Paulina, Juan, Daira Emma

    • A three-person team from Runtime Verification proposed formally verifying Zebra's arithmetic and parsing layer (zebra-chain) in Lean.
    • The cryptographic primitives in this layer are already built on audited libraries, and while tests look at only some inputs, a proof verifies all inputs.
    • Rust code is translated into Lean with tools (such as Aeneas) to prove lemmas and specification theorems.
    • The first four-week pilot proves at least 18 theorems.
    • The targets are amount arithmetic, non-canonical encoding of compact size, and height computation, and such bugs can cause consensus divergence between nodes.
    • The full roadmap is four to six months in four phases.
    • It needs two to three hours of confirmation from core developers during the pilot, and the results are kept in a public repository where anyone can re-verify them with the Lean kernel.
    • Daira Emma said the timeline is long, since AI has sped up formal verification.
    • They said that recent real bugs were mostly in elliptic curve point parsing (low-order points, the identity, non-canonical encoding), so these should be moved earlier in the schedule, and they asked for collaboration so that it fits with other formal verification work.
    • Natalie answered that the roadmap can be redrawn.
    • Daira Emma asked whether the formalization could keep up with changes to the Rust code, and Juan answered that it should be possible with AI assistance and CI integration.
    • Pili said the ZF team's grant review would be around next week.
  11. Open discussion: Crosslink testnet forks and the rollback limit

    Daira Emma, Nate

    • Daira Emma asked why a level of fragmentation that did not occur on the private testnet arose.
    • Nate answered that several causes overlapped, including a PoW block sync bug or performance problem in the new networking protocol, and a setting that made nodes CPU mine by default so people could get CTA without a faucet.
    • Daira Emma said that Zebra does not roll back more than 100 blocks and stays on its own chain, making recovery difficult, while zcashd shuts down, and that Zebra should also raise its rollback limit to 600 blocks and change its behavior.
    • Nate countered that it may be safer for a node to warn loudly and halt on a rollback.
    • Daira Emma explained that in Crosslink the finality point is determined objectively, so only what is finalized needs to be stored permanently, and fork depth warnings become a matter of the operator's local configuration.
    • Nate said that instead of Zebra's fixed limit, it should roll back dynamically to the most recent finalized point, and that this is not yet implemented.
    • Nate stressed that Crosslink turns the subjective guess of whether there is a PoW fork into the locally decidable objective fact of either a PoW fork or a finality stall.
  12. Open presentation: ZecD server-side wallet open source

    Zec.rocks / Overpay · Blake, Daira Emma

    • Blake had been building their own server-side wallet for Overpay, which lets you buy anything with Zcash, and released it as open source under the name ZecD on the Zec.rocks GitHub.
    • It does not aim for backward compatibility with zcashd and follows the Bitcoin RPC pattern.
    • It is stateless so all funds can be recovered with a seed phrase, addresses are derived from the seed, and only account 0 is supported.
    • It talks directly only to Zebra, has a regtest test suite, and targets a scale of millions of addresses.
    • The only migration path from zcashd to ZecD is to send funds on-chain.
    • Transparent address support is being released that day as a branch, and a security audit is being pursued for the shielded scope.
    • Daira Emma welcomed it, saying Zallet also needs competition, and said that while it is a pity about the compromise of supporting transparent addresses, they understand the practical judgment.
  13. Open discussion: proposal to support zcashd after NU 6.3

    Daira Emma, Jason, Pili

    • Daira Emma said they had heard that Shielded Labs proposed supporting zcashd even after NU 6.3 activation, and that they think it is not a good idea.
    • The reason is that core developers will decide on consensus bug fixes without considering zcashd, and if only Zebra is the supported implementation it is easier to ensure consistency with the specification, but if zcashd remains, every divergence becomes a security bug that must be addressed immediately.
    • Jason said the goal is the same: activate Ironwood quickly and have operators move safely to Zebra.
    • For mining pools, exchanges and others that cannot keep up with the current aggressive schedule, they said they are exploring a short-term bridge in which the Zero stack that Shielded Labs is building temporarily provides zcashd compatibility to help them get up to Ironwood.
    • Some mining pools say the schedule is too fast because Zallet is alpha.
    • Daira Emma said that if major mining pools use zcashd there is a risk that consensus divergences between zcashd and Zebra, which AI is finding, will remain for months, and suggested reconsidering the NU 6.3 schedule rather than deprecating zcashd.
    • Pili wrapped up by saying there was no time and suggesting to continue the discussion offline.

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