Nimbus
-
Mainnet launch, documentation
-
Thanks for supporters, in particular those who supported us on Gitcoin over a year ago.
-
Hot fixes yesterday, another scheduled today
-
Release management: mailing lists, hash of releases
-
Software:
- import/export using the v5 of interchange format WIP
-
Networking:
- Raised default peer requested to 160 from 79. Much better Attestation inclusion delay
- Asking for a networking call
- Seeing a drop of lighthouse clients over time
-
Cryptography
- 2.5x on square root implying 40% perf improvement on Hash-to-G2 -> This is 33% of verification so overall 15% perf improvement. -> supranational/blst#2 (comment)
Done
-> Fork choice, test vectors likely in January -> API testing: Protolambda looking for help
Testnet status? -> Pyrmont serves our purpose for the moment -> weak subjectivity sync
Nimbus
Lodestar
- BLST almost ready, will do bench of BLST WASM vs MCL WASM
- Cut a release just before mainnet
Prysm Done before mainnet Documentation
WIP now Upgrading the Slashing validator exchange format weak subjectivity sync eth2.0-API
Teku Done: weak subjectivity sync Video demo of Meredith, coming out soon Sync by backfilling block until genesis Race condition on Pyrmont (impact attestation not included) Issues with voluntary exits Pulling data from mainnet clients is harder (low powered hardware, Geth/Nethermind/OpenEthereum issues)
Lighthouse Not there "Mainnet running fine"
Nethermind - Tomasz For Eth1, Nethermind needs about 100GB of disk space
Justin: anyone working on removing the need of Eth1 node? Danny: anyone working on implementing a light protocol? Tomasz: (?) client might be able to go down to 40GB also working with Trinity on witness protocol to be even lighter
Aditya Weak subjectivity: in a branch now, weak subjectivity will depend on validators' balance Reworking on slashing protection: https://github.com/adiasg/eth2-slashing-protection-rebuild
Vitalik Light client PR #(?) Spec to write for
- incentives, accounting
- empty epochs so that long strings of empty epochs are costless
- inactivity leaks PR #(?), blog post coming
Proto: All clients are optimizing epoch transition in a similar manner and this acceleration data structure might directly be used in spec (and hashed)
Mikhail:
- Executable Beacon CHain has been published https://ethresear.ch/t/executable-beacon-chain/8271
Justin:
- BLST square root optimization 2.5x -> 40% perf improvement on hash-to-G2
Leo:
- Report written (behavior in non-finality, period of no blocks)
Tentative call wednesday
Vitalik available
Q: on state of phase 1 and phase 2 spec and tests -> low-hanging fruits? => data availability proofs => Kate commitments
Vitalik: fast linear combination = multiexponentiation
Justin: most low hanging fruit = light client PR of light client sync protocol in eth2.0-specs