Recovery during speculation #72
Description
The enhancement involves implementing a recovery scheme within hs that allows speculative execution to continue from where it was aborted instead of restarting. This could be due to network interruptions, user input cancellations, or other unforeseen issues.
Implementation Details
- State Logging: hs will maintain a log of the scheduler's state and the state of each command in execution. This log will be continuously updated to reflect the current execution environment. This shouldn't be very hard to implement since we already have a logger that does some command execution accounting.
- Recovery Trigger: Upon detecting an interruption (network issue, user cancellation, etc.), hs will trigger the recovery process.
- Resuming Execution: Using the logged state, hs will resume execution from the exact point of interruption, ensuring no loss of progress. We need to discuss further what state info is necessary for a recovery. We also need to determine what conditions should be fulfilled for successful recovery.
Next Steps
- More discussion is needed