tl.
- Advancement Progress: Minor Spec Upgrades and Full Steam Engineering 🚂
- No Innovation in Customer Diversity: Drive a Minor Client Instead of Being Greedy!
Merge Update
The engineering team did an excellent job. The Kintsugi Sprintwrap concluded in the launch of the Kintsugi Merge testnet. It was great to witness five execution clients and three consensus clients, in total comprising fifteen Pairs of clients working together as one.
The very first extended Merge testnet was met with great enthusiasm. The #TestTheFusion hashtag was trending on Twitter. The testnet was hit hard with transactions, faulty blocks, and a variety of other chaotic inputs. This resulted in some timing and state transition errors. We anticipated these issues in the early testnet launches. But with each iteration, clients become more stable.
Oven Reset 🔥🧱
The engineering team identified a major issue a few weeks ago. Engine API (the engine that powers the PoS consensus layer) had a mismatch in semantics related to the functioning of the execution layer clients. In brief, the consensus layer had inadvertently inflicted an unexpected load on the execution layer in some cases.
The engineers then realized that the API semantics could be made more flexible to enable the two layers to work together better. This resulted in a subtle, yet essential, alteration to the Engine API related rip specific release.
Today the engineers are working to reverse the alterations. The teams aim to have production-ready deployments on a fresh testnet by the end of the sprint. Stay tuned for details on how to participate.
After that, the teams will move on from public testnets to proof-of-stake and begin preparations for mainnet.
Customer Diversity Metrics
Michael Sproul recently proposed a new set of customer diversity metrics. Its unique fingerprinting mechanism is being utilized. However, the distribution of validator nodes has not changed in the past six months.
The diversity of client implementations on the consensus layer provides Ethereum users with a unique resilience against software glitches or attacks. A minority customer can give you some level of resilience irrespective of your network composition. But the network also gains resilience when it attains certain thresholds for validator distribution.
If one customer:
- Does not exceed 66.6%. A failure or error in one client is not considered a failure. Can’t go on
- Does not exceed 50% for any single customer’s fork selection error/failure. You can’t control all aspects
