# Week 16 & 17
This weeks we focused on fixing our eip-7928 implementation based on the current specs modification as well as running kurtosis.
## Changes made for BAL
* A particular test case focusing on `NonceChange` brought into our attention that balance change for an account is stored incorrectly. This was chiefly because we were fetching the pre and post balance directly from revm through `transfer` or `balance_incr`.So if an account's balance gets modified more than once in a block,the last transition was only getting stored.
To fix this, I fetched the balance concerning an address from database and compared it to the uncommitted state.This mitigated the problem.

* Another case which was discussed in the latest breakout was `self-transfer`. We had the initial setup for this,it only lacked checking `tx().to()` is same as `tx().signer()`.I added this too.

* `EIP-7702 Delegations: The authority address MUST be included with the nonce and code changes after any successful delegation set, reset, or update, and MUST also be included with an empty change set if authorization fails due to an invalid nonce. The delegation target MUST NOT be included during delegation creation and MUST be included when loaded as a call target under authority execution.`
This was majorly handled by my friend,I added nonce and code change tracking for authlist in pre-execution.

* For `7702 Delegation Clear`test case,we had one significiant problem.For `KECCAK_EMPTY` code_hash concerning an account we were using `Bytecode::default().bytecode()` as the code/code change(also for generalized cases).I modified this to `Bytes::fefault()` for code change and also added a bool to the `CodeChange` of `Account` in revm just to keep in track if a code change is performed or not. So accessing on evm became pretty straightforward i.e. checking whether `!code_change.is_empty()||modified`.
* Resolving the leftover errors,we got **88/90** passes in bal branch of reth. The ones failing is due to extra account entry for `Call And Oog` and `Delegatecall And Oog` for oog cases by eest.This will be fixed soon I believe :)
* I also tried running kurtosis with [jihoon's setup](https://github.com/jihoonsong/local-devnet-bals) before fixing the tests.I am due running one again with the fresh batch which I will do soon.

## Additional Information
Apart from that I attended the Bal Breakout and reported our present progress.The next set of discussion on it will be Coinbase and its underlying BAL incorporation complexity.
I also attended the weekly standup on Monday and office hours on Wednesday.They were very much insightful.
Dragan has refactored our original work in revm to make the implementation less heavy on alloy/evm.This will make the work all the more better.
Also much of week got spent solving problems concerning booking for attending devcon.Thanks to Josh and Mario for saving the day!!!

## Next Week Plan
* Incorporate Dragan's branch into BAL and add other modifications.
* Run kurtosis and share some **good** logs.

## Resources
*[bal-devnet-0 branch from reth](https://github.com/paradigmxyz/reth/tree/bal-devnet-0)
*[test result from present eest release](https://github.com/paradigmxyz/reth/actions/runs/18440365192/job/52540339783?pr=18432)
*[bal branch in evm](https://github.com/Rimeeeeee/evm/tree/bal)
*[bal branch in revm](https://github.com/Soubhik-10/revm/tree/bal)
*[test result in pokebal](https://pokebal.raxhvl.com/)