description: Notes from the regular Eth2 implementers call
# Eth2 Implementers’ Call #31 - 2020-01-09
[Quick contemporaneous notes by Ben Edgington]
## Testing and Release updates
Working on v0.10 spec. This will be a stable release, both for testnets and audit purposes. To be released by tomorrow. Contains breaking BLS changes, small modifications to network spec, configuration parameters, and other generally non-breaking changes.
No big updates on testing. Focusing now on testing for Ph1.
Testing for fork choice? [Mikahail/Harmony] Published a proposal on this: https://github.com/ethereum/eth2.0-spec-tests/issues/17 [Danny] With a couple of approvals from the client teams we can integrate this as well. **All, please review**
## Client updates
Up to 0.9.4. For fuzzing, keeping a separate v0.9.3 branch. Focused on networking: first PR on discv5 in place; verifying interop between _nim-libp2p_ and _go-libp2p_. Reworked sync; RPC endpoint.
Eth1 data management, including polling for block data. Syncing making progress. Disc v5 integrated - based on the Harmony module. Beacon-fuzz integration almost complete. New BLS hash-to-curve merged.
Q1. Any resources on how clients are syncing up with Eth1 main chain?
[Danny] Main approach: talk to an Eth1 RPC endpoint, assume it's a trusted node. There is a change in v0.10 to use timestamping rather than block height, which makes caching easier.
Q2. For beacon fuzz, currently at 0.9.1 - what's the plan for updating?
[Sorry, missed some of this!] Working on spec v0.9.3; will move to v0.9.4 soon.
Re-launching [testnet](https://prylabs.net/participate) today :tada:. Uses mainnet config with 3.2 Eth deposit size. Up to v0.9.3 spec. Needed some significant optimisations to run mainnet spec. Performs attestation aggregation.
Testnet has been running again, will announce publicly soon. Mainnet spec, and running stably for a couple of weeks. Close to a version 1.1 release of the client. Removed requirent for a full Eth1 node - can use a fast-sync node. Built docker-compose infrastructure. Added Herumi BLS - 2x speed gains. Adding naive aggregation strategy - will put this on to a new testnet.
At spec v0.9.4. Working on discv5 simulation. Discussing time consensus, working on a write-up of this.
Final stages of merging 0.9 branch into master: v0.9.3. Working on refactoring SSZ to allow for re-merkleable library.
[Protolambda] [remerkleable](https://github.com/protolambda/remerkleable): SSZ objects can be very large. Some kind of data sharing in the library is desirable. Act like it is a normal spec object, the library takes care of all hashing and serialisation in the background.
Block signing working. At v0.9.1. Can now run a chain. No attestations yet; mock Eth1 data. Working on peering next. Plan to add some more people to the team soon.
## Research Updates
Mostly Ph1 and Ph2. Some improvements to data availability constructions. Will start with 2D erasure coding. There may be better constructions in future.
Ph2, some ideas around how to do execution environments.
Proposal for moving Eth1 into Eth2 on an earlier schedule than previously envisaged. Prerequisite is getting stateless clients working.
Supranational is working on VDF ASIC. They are also looking at a way to accelerate BLS12-381 for CPUs. Point addition is already 10% faster than Herumi. They believe formal verification of Herumi may be difficult; they are taking a different approach more suited to formal verification.
Working on acceleration of Snarks. ASIC could offer 1000x proving performance improvement.
Another VDF day in planning, 1 day prior to SBC, on 18 February.
Phase 2 call on Tuesday 14th.
There is now a Ganache-like tool that can run EEs on multiple shards. Will release a roadmap and doc this week.
Working with TXRX team to build end-to-end testing suite. Also will have roadmap doc soon.
Overlap with Eth1 stateless work. Will publish an overview and recommendation for an approach. Starting a prototype based on actual Eth1 data.
Continuing to look at some cross-shard schemes. Other write-ups coming: EE upgradability; fee markets for Ph1; some interim iterative plans for Ph2.
Collaborating with Quilt on cross-shard Tx simulator.
Anomlee ML EE prototype - a trained neural network.
Also ethsear.ch <- an Ethereum-dedicated search engine.
Gathering speakers for [Eth Barcelona](https://ethbarcelona.github.io/) - contact him if you are interested.
Streamed network responses have made it into v0.10. There will be another networking call in 1 or 2 weeks.
## Spec discussion
[Vitalik] How do people feel about an accelerated Eth1 -> Eth2 switch?
From technical point of view, it's good. Will the community accept it? Depends on timing, and successful Ph1 and Ph2 launches and running.
Will it replace finality gadget integration? FG could be implemented voluntarily before a switchover, but basically this is an alternative to a protocol-level integration of the finality gadget.
There's quite a lot of work to do on 1.x stateless implementation before this. It's a huge priority this year.
Other update: formal verification of deposit contract - is going deeper than usual due to question marks about the quality of the Vyper compiler.
## Open discussion
Contact Danny if you will be at Eth Denver!
* * *
# Chat highlights
From danny to Everyone: 02:03 PM
From Mikhail Kalinin to Everyone: 02:08 PM
From Terence to Everyone: 02:18 PM
Genesis time: 2020–01–10 T00:02:00.000Z
From danny to Everyone: 02:23 PM
From danny to Everyone: 02:40 PM