What's New in Eth2 - 20 September 2019

Edition 26. Archive. Trouble viewing? Load original.

Ben Edgington (PegaSys, ConsenSys — but views expressed are all my own)

The Interop Edition

In many ways it's been unusually quiet on the Eth2 news front these last three weeks. But the reasons for that are significant: everybody has been totally focused on the Interop Lock-in event.

The Interop Lock-in

For those who haven't been paying attention, last week, all but one of the Ethereum 2.0 client development teams gathered in remote Ontario with the goal of getting their beacon chain nodes talking to each other. We were joined by Ethereum Foundation people, Whiteblock people, and Eth2 Phase 2 people (the ConsenSys Quilt team and the EF Ewasm team).

I'm not going to rehash the whole event here. Danny Ryan has published a review on the EF blog, and I've got a write-up coming for the ConsenSys blog in a couple of days — I'll tweet it out for y'all when it's there. Suffice it to say that the event was successful beyond anything that we had expected beforehand.

Some highlight tweets:

• #Eth2Interop is 🔥
• Lighthouse and Nimbus declare interoperability
• Artemis and Nimbus are next
• Lodestar and Lighthouse
• Prysm and Lighthouse
• Artemis and Trinity
• Trinity and Lodestar
• Four way interop: ArtHouseBusStar
• Five way: ArtPryHouseBusIty
• And the pièce de résistance, 7-way interoperability!
• A special mention for Wei Tang of Parity. He couldn't be on site, but was still able to get Shasper talking to Lighthouse

Other tweetable moments:

• Talking chain syncing
• Nimbus and Lighthouse running together on a two-node Raspberry Pi network
• Greg the Greek looks sceptical
• People fell in the water
• People tried to walk through glass doors.
• My reflections from Toronto airport on the way home

I've made an album of a few of my pictures.

Implementers' call

Call #25 took place on the 19th of September.

• Agenda
• Video
• My quick notes and Pooja Ranjan's thorough notes.

It was a quick one — only 45 minutes. I think everyone is exhausted after Interop, and taking a breather before DevCon.

Notable: spec version 0.8.4 will be out in the next days. This will have updates to the networking specification, and some special test vectors to cover a small number consensus failures teams experienced during Interop. The core spec remains frozen, and no updates are anticipated.

BLS Standardisation

One big topic that hasn't been much discussed is the standardisation effort for the BLS signature scheme used by Ethereum 2.0. Several next-generation blockchains (e.g. Algorand, Chia, Dfinity) plan to use BLS signatures on the BLS12-381 curve, like us, and it makes sense to adopt a common approach to facilitate interoperability in future.

Now, the current implementation of BLS signatures that Eth2 clients are using differs from the proposed standard. In particular, with respect to the hash-to-curve algorithm (this is how data to be signed is mapped to a point on the elliptic curve).

Kirk Baird of Sigma Prime has started drafting an update to the Eth2 BLS specification. The current method we use to perform hash-to-curve (the "try and increment" technique) is pleasantly simple. Unfortunately it doesn't make it into the standard, largely because it is not constant-time. Constant-time is not really a feature that we need in Eth2 — all the input information to the hashing is public. Nevertheless, we need to move to the new algorithm to comply with the new hash to curve standard.

For reference, I've made an implementation of the new scheme in Java that passes all the test vectors. Kirk is working on an implementation in Python. My implementation is about 50x more lines compared with our current Java version, excluding tests 😂.

A significant consequence of all of this is that the deployment of the Validator Deposit contract on the Ethereum 1 chain will likely be delayed. The original plan was to deploy this in a ceremony at DevCon — this looks unlikely now. The issue is that part of the validator registration process involves generating a BLS signature: until the standard is frozen, we can't be sure that this process won't change. Word is that we should, however, be ready to go in a few weeks, and this shouldn't delay the planned start of the beacon chain in Q1 next year.

Research

A couple of things, briefly.

Vitalik has some thoughts about a different way Phase 2 execution could be implemented. This was intensively discussed by the Phase 2 teams during Interop. I need to digest it more before opining on it here.

Ryuya Nakamura has continued thinking about attacks on Casper FFG. He's followed up his work on the Flip-flop Attack with an Analysis of bouncing attack on FFG, and a follow-up, Prevention of bouncing attack on FFG.

In other news

Sorry if I've missed your thing below; it's been a bit intense lately.

• Despite Interop, Lighthouse and Prysm got client updates out.
• Slides from Lighthouse's Mehdi at EthBerlin on Performance & Security for Ethereum 2.0
• David Hoffman live-tweeted Vitalik from Ethereal Tel Aviv. Some good Eth2 stuff in there.
• Colin Schwartz continues his "Ethereum 2.0: A Complete Guide" with Ewasm.
• A Casper FFG explainer from Juin Chiu.
• Did I mention that I've registered the domains eth2.news and eth2.info? The first points to What's New in Eth2 (i.e. here). The second to a WIP collection of what I believe to be some of the better resources for getting up to speed on Eth2 for newcomers.
• This week's comedy reward for reading this far: Joe Lubin and Vitalik Just Keep Chugging.

Next stop, DevCon! Come and give me the secret handshake if you spot me around Osaka 🤝 またね

Follow me on Twitter and/or Peepeth to hear when the next edition is out 🙌.

We also have an RSS feed.