ACDE 190 Recap

Original post:

We wrapped up another @ethereum #ACDE early today! Everyone is heads down working on Pectra.


The first thing on the call, we discussed the pectra-devnet-1 spec:

It's pretty close to devnet-0, with several bug fixes found during interop, the addition of EL-triggered withdrawals, and swapping EIP-3074 for 7702.

EL teams were all still working on their implementations, with 7702 understandably being least far along across the board. Besu had some gas pricing questions on this EIP, Nethermind and Geth are still rebasing/merging various branches for the EIPs.

Erigon also needs to implement the EL-triggered withdrawals. Reth and EthJS were both nearly ready, with both teams missing 7702.

We then spent a large chunk of the call discussing that EIP, and specifically how to approach revocations. This is something we still haven't reached broad consensus on. We briefly covered a first proposal by @frangio_:

One concern with this, raised by @lightclients, is that it can be implemented at the contract level, and that this may not be the right thing to implement in the protocol directly.

@lightclients had his own proposal for a delegation scheme, where a delegated address is explicitly set as delegator (and can be revoked by redelegating to, say, 0x0):

There was a bit of pushback on the call by wallet developers, saying this would make certain use cases more complex, or impossible. On the other hand, while client teams had not fully reviewed the proposal (the PR was <24h old!), there was interest in exploring this approach further.

One thing I tried emphasizing on the call is that the core tradeoff isn't as much "this vs. a perfect solution" but "something that improves user + dev experience in Pectra vs. nothing in the fork."

While we obviously should ship something that is useful, it's probably fine to start with something more restrictive and see if we want to relax some constraints in the future.

And UX vs. security is a really hard tradeoff. For example, this proposal would require 1 signature per chain. Some wallet developers highlighted this is bad UX, but on the other hand, it also means you cannot lose your assets across all chains with a single signature...!

But, again, this was a pretty new proposal and there are further discussions to be had. We tabled the topic and agreed to have another breakout on 7702 before the next ACDE. We're currently polling for the best times 😄

If 7702/AA is something you have a strong opinion about, I recommend watching the recording for the entire back and forth and monitoring for the breakout info!

Next up, we discussed progress on EOF devnets. @BarnabasBusa reported that on the EOF breakout, teams had decided to base the EOF devnets on Pectra, which means a lot of the work needs to be rebased before the devnet goes live.

This is a bit more up-front work, and may make it harder to distinguish between devnet-1 bugs vs. EOF bugs, but that is something we'd have to tackle at some point anyways, so better sooner than later 😄

Progress-wise, things are being tracked here:

Besu is good, Nethermind is almost ready, with their implementation still failing some tests, and same for Geth. Erigon was passing all static tests and would now like to test against full EOF blocks.

Reth is passing tests and work has begun towards integrating EOF in Foundry!

We're still a few weeks from a full devnet, but things are coming along nicely enough that the EOF breakouts will now happen every 2 weeks vs. weekly 😄

We then briefly touched on PeerDAS progress: the next devnet will be based on the recent CL spec release (v1.5.0-alpha.3) and is expected in the next couple of weeks. More on that in next week's CL call!

And, to wrap up, we had three quick announcements. First up, I proposed a discourse template for EIP discussion-to threads:

These would make it easy to see the current state of EIPs and track high-quality commentary. No objections on ACD! Template:

After, @parithosh_j pointed out that only has CL client dev admins and asked whether EL devs also wanted to be admins on the repo.

And, lastly, we agreed that ACD will happen as scheduled in two weeks, on July 4th. I'll be OOO, but @ralexstokes will be there to run things. I'll be back running my next ACDE on July 18, 14:00 UTC!

Collect this post to permanently own it.
Subscribe to timbeiko.eth and never miss a post.