### Dencun testing call notes by [Terence](https://twitter.com/terencechain)

_Copied from [Eth R&D Discord](https://discord.com/channels/595666850260713488/1031999860997619843/1153366840819601500)_

**Devnet8 Summary:**
- Issues have been identified with validator keys running on different machines. This is not the intended behavior. Ethdevops is currently investigating the problem.
- There was good finalization over the past week.
- Both Besu and Prysm had issues, but they have since been addressed.
- Erigon has an ongoing issue with returning the incorrect execution API status. The issue is due to a transient error where, in the event of running out of memory, it shouldn't return as invalid. For Erigon users, refer to [this commit](https://github.com/ledgerwatch/erigon/commit/751de6a0287b7f86308078a5fd438d3d23f5298d) for devnet8.
- Blob spamming is currently ongoing, and no issues have been observed.
- Devnet9 will introduce a new blob spamming feature.
- Lighthouse is currently addressing a single blob lookup race. It aims to reduce the number of lookups it's making as it's currently over-querying. There's also an issue with incorrect caching, especially in scenarios where a blob is valid on its own but later on, the kzg commitment of the block doesn't match the blob. These test cases will be added to devnet9.
- Prysm besu is experiencing slow syncing. This will be reviewed subsequently.

---

**Devnet9 Updates for Execution Spec Test and Hive:**
- **Execution Spec Tests:**
- EIP-1153 tests have been merged into the main branch.
- EIP-4788 PR will be updated to its latest version. The deployment transaction is the only thing missing from the latest EIP PR. Once that's added, it will be merged and released.
- EIP-7516 tests will soon be merged and included in the next release.
- **Hive:**
- There is a test case for the Geth London non-zero-block-number issue.
- The mock builder is now Deneb-ready and can introduce errors.

---

**Discussion on Empty Accounts:**
- Empty accounts are remnants from an older network. Such accounts are unlikely to occur in production settings.
- The consensus is to assert that a merge can't happen if there's an empty account present in the network. Furthermore, clients aren't mandated to support these empty accounts.
- One suggestion was to specify that genesis allocations cannot define empty accounts. Additionally, execution spec tests should be modified to exclude empty accounts in the genesis.
- There's a need for a reliable process to review all the accounts post-merge to ensure none are empty.
- The majority of the discussion on this topic can be found [here](https://notes.ethereum.org/@RG8j1GXNQ-iPkkFqXEouVA/By-ep0ry6). In summary: Empty accounts won't be allowed in genesis, there's no standard specification for this, each client has its unique specification, and this rule will be enforced in the hive test.

---

**Devnet9 Status:**
- Two new features have been introduced: activation churn and blob gas base fee.
- The aim is to have these features implemented by Thursday. Afterward, a decision will be made on the next devnet date. Further updates will be provided in due course.