← All calls
ARBORIST#112

Zebra 3 RC cuts sync time sharply

The Zebra 3 release candidate is out, and a full sync now takes 8 to 16 hours.

Agenda summary 8

  1. Zebra update

    Zcash Foundation · Marek

    • The Zebra 3 RC, the release candidate for NU 6.1, was released and will be finalized as Zebra 3 if no problems appear.
    • Many external-contributor PRs on error handling, the RPC API and regtest mode were merged, and differences between Zebra regtest and zcashd reported by the Zaino team were fixed.
    • Thanks to performance improvements, Zebra 3 syncs much faster than Zebra 2.
    • The fastest sync took 8 hours 30 minutes and the slowest about 16 hours, compared with 50 to 70 hours before.
    • The build documentation now says that installing RocksDB as a system library and linking it dynamically avoids recompilation.
    • The migration of documentation hosting is not finished, so it is not yet rendered on the web.
  2. ECC core update

    Electric Coin Company · Kris

    • The NU 6.1 zcashd release is done, and because new Zashi users surged, the past two weeks went into fixing wallet-backend bugs.
    • The team said the number of new Zashi users added in the past two weeks is similar to the growth of the whole past year.
    • Str4d opened a PR adding password authentication to the Zallet API, and cookie authentication is in progress.
    • A new zcash_keys crate release was made, and releases of the rest of the librustzcash stack, including zcash_client_backend, are due around next week.
    • Daira Emma and Str4d are devoted to reviewing Qedit's Orchard crate, so work on Zallet and zcashd retirement resumes in earnest after the next Arborist call.
  3. Zaino update

    Zingo Labs · Dorian

    • Four RPC endpoints were completed, one of them a contribution from Pacu.
    • Two are under review.
    • On the Zallet side, work continues on integrating the chain index to replace the existing local cache.
  4. zcashd retirement update

    Electric Coin Company, Zcash Foundation, Zingo Labs · Pacu

    • Partner responses rose sharply because of growing interest in Zcash.
    • Interest grew in the friction survey on the burden of the network upgrade process, including transaction changes, as well as in the Zallet alpha, and from block explorers and API providers.
    • Work to retire lightwalletd has begun.
    • This overlaps with the light client working group: adding transparent data to the protocol, for shielded and transparent support in mobile wallets and light clients, reduces what servers can learn.
  5. ZSA update

    Qedit · Vivek

    • Most ZIP changes were finished last month and are stable; a PR refactoring the Orchard note plaintext to fit both ZSA and the memo bundle is open in the zips repository.
    • The upstream Orchard PR is ready so it can be reviewed as soon as ECC wants.
    • After looking over the changes, Daira Emma pointed out a small information leak about which assets are not being transferred, and the fix of shuffling the order of actions has already been made.
    • librustzcash is aligned with the ZIP changes and upstream; a draft of the ZSA fee implementation has been written and tests are being added.
    • This PR is expected to be reviewed after the Orchard review.
    • The Zebra fork has caught up to 2.4.2 and made progress on the additional state management ZSA needs.
    • It also works with the latest Orchard and librustzcash, and the test vectors reflect the ZIP changes.
  6. Network Sustainability Mechanism (NSM) update

    Shielded Labs · Mark

    • Mariusz is keeping the two PRs posted in chat up to date with main, and they are ready to merge after a few more approvals.
    • After merging, work moves on to ZIP 234 and ZIP 235.
    • Kris again asked Daira Emma to review the test vectors.
  7. Crosslink update (Trailing Finality Layer)

    Shielded Labs, Electric Coin Company · Andrew, Nate

    • A demo video showed that simply running the executable with no configuration joins and syncs with the existing testnet and receives both PoW and PoS blocks.
    • Roster management is temporarily stored in PoW block headers, so even running only the built-in miner gets onto the roster without manual intervention.
    • The header approach has a longer propagation delay than the mempool, which creates local side chains and requires requests to be resubmitted; it is not the production approach.
    • Staking on an address implicitly makes it a roster member, and once finalized, rewards are distributed in proportion to stake.
    • A workshop will be held on Wednesday the 22nd; building with the default settings and running is enough to join the network, which is easier than last time.
    • The team is considering ways to scale the testnet to production level and to encourage participation with real ZEC, such as earning testnet staking eligibility by burning mainnet ZEC.
    • An issue (number 24) was opened to prepare in advance a permissionless emergency patch that turns PoS off and returns to PoW.
    • The idea is that being able to respond with an emergency upgrade instead of a freeze button helps keep the unbonding period short.
    • An emergency patch is effectively an emergency network upgrade, so most of the mining power would need to install the new node, and a chain split is possible.
    • As a precaution, the effect of the relay attack protection introduced by Overwinter was mentioned.
  8. Open discussion

    Nate, Dorian

    • Nate said they want to standardize burning ZEC and carrying public information in transactions for app developers.
    • zecmec, who is designing a social app that uses ZEC burning to eliminate spam, is looking for people to help code the backend wallet.
    • Dorian wrote a specification for endpoints that depend on zcashd implementation details and will not be implemented in Zebra, and shared it in the Zaino Discord channel.

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