Feedback from the Besu team:
- Some EIPs were not specced sufficiently and delayed the fork because several rounds of spec changes where necessary along the way
- Scoping Pectra took way too long. Freezing the SFI list should happen very early in the planning process
- EIP-7742 was dropped too quickly. It might make sense to revisit it in the future to allow CL only hard forks
- PandaOps and testing teams are top notch
- Gating devnets seems like a good idea, but wasn’t strictly enforced. Seems to still make sense to do it
- Hive tests > EEST tests. Hive uncovered more bugs in our case, because EEST does not execute100% of the code path. Hive tests should be used for gating devnets.