# Week 14 & 15 This week was by far most significant for our project as we managed to make it almost complete (as per the current state of specs) :) ## Work Update: We had very little time this week due to our uni exams. As I have mentioned in my earlier update, the eest tests were using `engine_newPayloadV4` instead of the proposed `engine_newPayloadV5`.However last week a new PR was merged into the eest that fixed the version.I also had my mid Cohort check-in ( which was awesome, ty Josh and Mario for all the support ). Before the call I generated the new fixture from the eest and started the tests.However to our dismay all the tests were failing which was odd as they had passed before. I and my friend investigated the new changes and found a new check was needed. Upon adding that we found that there were no changes recorded. But we did not have time as our exams were still going on. So after a couple of days our last exam got cancelled due to heavy rain and flooding in the city. It was devastating sure but gave us just the time we needed.We started working immediately and rebased evm and reth repo.There was some issues while rebasing.I added bunch of traces and inspected the logs. **But there were few mismatches, namely one address was missing from every BAL and two BAL had problematic entries**. ## Bugs and their fixes First was the nonce change which caused few problems having the same pre and post balance as we donot keep the exact pre balance, we keep the previous balance before the last change. So I fixed it by giving a new flag that tracks exactly that. And the address that was missing from every block was the coinbase. So we both tried fixing it in many ways but it did not work. Ultimately I fixed it by just checking it after each txn. And the final bug was the code change recording as we did not have the contract address deployed by the `CREATE` call from a factory. Recently I refactored all these changes to simply fetch from the result and state so its compact and easy to read. With all the modifications, the consume-rlp tests were passing but engine were failing. We investigated the logs and found **block hash mismatch** and **invalid payloads**. We both added a bunch of traces but nothing worked. Finally we decided to reread our implementation, I read the reth and Soubhik the alloy changes. I fixed couple of message version mismatches but it was still failing. There was an error mapping issue. Upon fixing that, [it was all green](https://github.com/paradigmxyz/reth/actions/runs/18055613513/job/51385001065?pr=18432). ## Joining the Breakout I also joined the Bal Breakout with Dragan from reth, there was discussion about adding the precompiles addresses and some more changes concerning self destruction. Overall it was a very busy and important week. Next week we hope to tackle any new specs change and possibly try interop with other clients. ## Resources * [Failed consume-rlp job in reth branch](https://github.com/paradigmxyz/reth/actions/runs/17948674832/job/51042508762) * [Main BAL PR in reth](https://github.com/paradigmxyz/reth/pull/18432) * [Modification to evm impl](https://github.com/Rimeeeeee/evm/pull/8) * [Added missing conversion field in alloy](https://github.com/Soubhik-10/alloy/pull/15) * [Main branch for BAL in revm](https://github.com/Soubhik-10/revm/pull/19)