← All calls
ARBORIST#84

Crosslink v1 requirements published

zcashd 5.10.0-rc1 came out, and Zooko presented the Crosslink v1 requirements.

Agenda summary 15

  1. ECC update: zcashd 5.10.0-rc1 and transparent wallet history

    ECC · Kris

    • The zcashd 5.10.0-rc1 release PR went up this morning, making it possible to begin trials on the NU6 private testnet.
    • The main cause of delay was the aging of the RPC tests.
    • In NU6, miners are prohibited from not claiming all fees so blocks always balance exactly, and this also helped find a bug in the lockbox implementation.
    • The Zcash client backend and the SQLite backend now support full transparent wallet history recovery, which is an important step toward replacing the zcashd wallet.
    • Greg Pfeil opened a PR for a Rust interface to zcash_script's transparent scripts and began comparing the C++ and Rust interpreters with property tests.
    • They asked ZF to review that PR.
  2. Discussion of ZSF deposits and unclaimed rewards

    Tomek, Daira Emma, Kris, Paul Dann

    • Tomek pointed out that it is unclear in ZIP 233 whether past unclaimed rewards go to the ZSF.
    • Daira Emma answered that the ZSF deposit field will be an explicit field and that unclaimed rewards will be 0 going forward.
    • Kris said that if fees and deposits are implicit, verification on hardware wallets is hard, so a rule that blocks and transactions balance exactly is better.
    • Paul Dann said that if the ZSF is defined as the difference between the maximum and the current chain value, past unclaimed fees are also included automatically.
  3. NU6 testnet activation plan

    ECC, Zcash Foundation · Kris, Daira Emma, Pacu, Dodger

    • ECC and ZF engineers plan to test NU6 activation and mining of blocks before and after activation on the private testnet that day, and then set the public testnet activation height.
    • A dedicated Discord channel already exists.
    • Kris said zcashd 5.9.1's end-of-service halt is in early September so activating after that is safe, but Daira Emma countered that this policy does not apply to testnet so it is not necessarily required.
    • Pacu said infrastructure operators who are not taking part in testing need advance notice, and offered to help directly.
    • Kris promised to document the private testnet process as they go.
  4. Zebra update: Zebra 1.9 and RPC changes

    Zcash Foundation · Arya, Alfredo

    • They merged error handling in the block subsidy calculation to prevent a panic at low block heights, and made the coinbase balance exactly after NU6.
    • Even when the DB major version is bumped, the state cache can be restored from the previous version, so the chain does not need to be resynced from scratch.
    • The getblocksubsidy RPC returns lockbox stream information when NU6 is active, and getblockchaininfo returns the pool balance at the chain tip.
    • They decided not to put per-block pool balances in getblock, since they are not stored.
    • They opened a draft PR in which sendrawtransaction returns the reason for a validation failure, but because the lightwalletd change has not been merged it is not propagated.
    • Zebra 1.9 is ready to deploy the NU6 testnet.
  5. Security fix backports and ECC SDK and Zashi releases

    ECC, Zcash Foundation · Alfredo, Daira Emma, Kris

    • Alfredo gave their initial conclusion that Zebra is not affected by the two denial-of-service fixes backported from Bitcoin to zcashd.
    • zcashd 5.10.0-rc1 includes NU6 and these backports.
    • ECC released the Android and iOS SDKs and Zashi.
    • The new release looks up the median of several exchange rates over Tor and supports ZIP 320 TEX addresses.
    • Full transparent history recovery is work whose priority was moved up after its necessity became apparent during TEX support.
  6. zcashd deprecation and RPC tests

    ECC, Zcash Foundation · Alfredo, Kris, Pacu

    • They started an attempt to run zcashd's RPC tests with a Zebra backend and immediately found several issues, and decided to use this as the path for the deprecation work.
    • Kris said that many tests are tightly bound to zcashd behavior and will fail, but that it lets them compare the two nodes' responses to the same RPC request side by side.
    • Pacu said their grant goals were adjusted and extended by a year, and that they will take on anything they can help with in the zcashd deprecation.
  7. Zcash Sustainability Fund (ZSF) update

    Tomek, Paul Dann, Jason McGee, Daira Emma

    • The Zebra code has been nearly complete for about a month but has not been posted because integration testing is difficult, and there were problems in testing ZSF deposits on a zcashd and Zebra testnet.
    • They plan to push the librustzcash fork changes soon.
    • Paul Dann implemented ZSF in zcashd and successfully sent a transaction, and the coinbase deposit and some mining RPC updates remain.
    • Jason McGee is doing the zcashd implementation to get ZSF into the Q1 2025 upgrade, and said that if the deprecation slips to the second half they will push for a standalone upgrade.
    • Daira Emma said they want to avoid an additional zcashd upgrade but it depends on timing, and that the core of the change is a transaction version change.
    • Tomek agreed to upload the current Zebra code.
  8. Zcash Shielded Assets (ZSA) update

    QEDIT · Vivek

    • They applied the refactoring that reverts from v7 to the v6 transaction format across librustzcash, Orchard, the test vectors and more, and the dependencies support vanilla Orchard and ZSA together.
    • They changed the Zebra fork to use QEDIT dependencies, and added NU7 placeholder changes and an arbitrary activation height on top of NU6.
    • They built a ZSA test environment that runs the Zcash transaction tool as a Docker image on a Zebra node, and it is under review.
    • They finished the Zcash note encryption changes reflecting Str4d's review.
    • They caught up on the ZIPs changes resulting from NU6 and the repository reorganization, and two or three open ZSA-related issues remain.
    • They plan to review the asset swap ZIP once more and then announce it is ready to merge.
  9. FROST update

    Conrado, Pacu, Daira Emma

    • They are adding encryption and authentication to the FROST server, and key management such as public key registration and identity binding is tricky.
    • They expected the server audit to be possible in roughly 3 to 4 weeks.
    • The frost crate 2.0.0 has not been published yet, awaiting further feedback on the RC; the key share refresh for trusted dealers is done, while the one for DKG is in progress.
    • They met with Gordian earlier and will meet again in September; their work is closer to share storage and backup so it is not blocking now, but cooperation could be tried.
    • Daira Emma proposed in a ZIP PR a way to generate the alpha value without relying entirely on the coordinator's RNG, and Conrado welcomed it and said they would update the ZIP.
    • A draft of how to generate the FROST wallet key tree is covered in a separate PR 883.
  10. Crosslink v1 overview

    Shielded Labs · Zooko, Daira Emma

    • Shielded Labs is organizing requirements and tradeoffs to speed up Crosslink deployment and plans to implement the simplest prototype soon.
    • The motivations for introducing hybrid PoS are three: increased demand for ZEC with reduced circulating supply, finality and bridge support, and a foundation for future improvements.
    • Zooko said they will share a list of goals such as staking rewards, running one's own validator, time to launch, safety, user experience and latency, to gather input.
    • Daira Emma corrected that the attacker share needed to break PoS is 34% if partitions are assumed, and Zooko revised the list to 34% to 67%.
    • They also added other attack surfaces to the list, such as implementation bugs, light client servers and proof systems.
    • The next step is to gather missing goals, analyze tradeoffs and decide the v1 scope.
  11. Shielded Labs hiring announcement

    Shielded Labs · Jason McGee

    • They said they are hiring a few intermediate-level Rust developers for Crosslink development.
    • Interested people can contact Jason McGee via Twitter, the forum or email.
  12. Midnight's Halo 2 recursion work

    Midnight (IOHK) · Josh Swihart

    • Josh Swihart relayed from a conversation with Eron Barak that Midnight, a Cardano-family project, has nearly finished adding Halo 2 recursion.
    • Midnight plans to finish in Q4 this year, go through a 3 to 4 month audit and then upstream the code, and they want to arrange a connection with the core teams.
  13. Open discussion: recursive proofs and past chain verification, and a ZCG grant review request

    ZCG · Jon, Daira Emma, Jason McGee, Alfredo, Arya

    • When Jon asked whether recursion could prove all past transactions so that one could bootstrap instantly, Daira Emma answered that the transparent protocol is very complex to prove, so it can be verified once and then used as an input through social consensus.
    • Orchard is easy to prove thanks to the same-curve cycle, but Sapling is hard because it needs non-native field arithmetic.
    • They mentioned that in the future Sapling could also block new deposits and allow only withdrawals to limit potential balance violations.
    • To accelerate the zcashd deprecation, Jason McGee asked for review of a $76,000 grant to implement Zebra RPCs (subscription to tip changes, subscription to mempool changes, and nullifier lookup by block hash and height).
    • Daira Emma agreed to take on the review, and asked for comments before ZCG's next meeting on September 2.
    • Arya plans to hold a scope discussion meeting the next day and will invite Pacu, Jason McGee and Daira Emma.
  14. Open discussion: Crosslink reward distribution and why PoW is kept

    Shielded Labs · Daira Emma, Zooko

    • Daira Emma pointed out that mining rewards must be split between PoW and staking, which is equivalent to an additional halving, and that the issue is whether the absolute size is sufficient.
    • Zooko gave their thinking of splitting rewards 50-50.
    • Zooko explained that with PoW each victim costs the thermodynamic cost of one block, while with PoS, if it is breached, unlimited double spends are possible with the same capital.
    • Daira Emma confirmed that this is one of the motivations for keeping PoW in Crosslink.
    • Daira Emma said that if the ZEC price rises, the PoW share of the allocation can be lowered to reduce environmental cost.
  15. Open discussion: small participants and top-heavy risk

    Shielded Labs · Pacu, Zooko, Daira Emma, Arya

    • Pacu asked whether Crosslink is an opportunity to draw small miners back in, and whether small delegations would harm security.
    • Zooko raised again the top-heavy problem, in which large sums move through ZSA and if the ZEC price plunges, the cost of attack becomes much smaller than the asset value.
    • It was also discussed that the economic analysis breaks down against attackers with unlimited resources such as governments.
    • In response to Arya's question, Zooko answered that there is not yet a Git repository for the exploration implementation and that they will post it on the R&D Discord once there is one.

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