A few days ago I came across footage of Zhuque-3’s first-stage landing, and what people were arguing about in the comments wasn’t whether it looked good or not, but whether this flight should count toward the official reusable-use ledger. In the livestream, the legs held up, but the acceptance and success counts were still working through another process. Watching those few seconds, I suddenly felt this was a lot like where I got stuck while porting Rusk 1.7: seeing one execution does not mean one settlement is acknowledged. #dusk
After Boreas, archival records began to retain and label contract events that had already been rolled back, while removing rollback-related staking updates from the formal index. At first I found it annoying and thought: if it’s invalid, why keep it? Later I had to admit I had been too obsessed with keeping things clean before. A contract may first emit traces of a transfer or staking, and then the block may revoke that execution before finalization. If the index wipes it out on the spot, afterwards it becomes impossible to explain why users saw that jump at the time; if it is left as-is without labeling, downstream systems may take the voided change as a completed credit. The archive exists for reference, while the official ledger only recognizes the final valid state. The process can be preserved, but the balance must not be contaminated by the process. @Dusk
Fixing the node does not mean the wallet and the accounting service are reading it the same way. I’ve personally been burned by this: I get happy first when I find the event name, only to realize later that the state never actually landed. So when I look at $DUSK , I’m not just counting new features; I want to see whether the same rolled-back transaction will be written into the same conclusion across all endpoints. Being able to keep the invalid past on record without letting it enter the balance may look insignificant, but it is the foundation that keeps settlement trustworthy. I’m still cautiously optimistic about this one: the rule is already visible in the code, but whether the entire chain follows it consistently still needs more watching. $BTC
After Boreas, archival records began to retain and label contract events that had already been rolled back, while removing rollback-related staking updates from the formal index. At first I found it annoying and thought: if it’s invalid, why keep it? Later I had to admit I had been too obsessed with keeping things clean before. A contract may first emit traces of a transfer or staking, and then the block may revoke that execution before finalization. If the index wipes it out on the spot, afterwards it becomes impossible to explain why users saw that jump at the time; if it is left as-is without labeling, downstream systems may take the voided change as a completed credit. The archive exists for reference, while the official ledger only recognizes the final valid state. The process can be preserved, but the balance must not be contaminated by the process. @Dusk
Fixing the node does not mean the wallet and the accounting service are reading it the same way. I’ve personally been burned by this: I get happy first when I find the event name, only to realize later that the state never actually landed. So when I look at $DUSK , I’m not just counting new features; I want to see whether the same rolled-back transaction will be written into the same conclusion across all endpoints. Being able to keep the invalid past on record without letting it enter the balance may look insignificant, but it is the foundation that keeps settlement trustworthy. I’m still cautiously optimistic about this one: the rule is already visible in the code, but whether the entire chain follows it consistently still needs more watching. $BTC
