hydromodpy.workflow.resume

Resume planner driven by the workflow_steps journal.

The runner asks ResumePlanner where it must restart after a crash or interruption. The planner reads the journal, verifies each completed step’s artefacts still exist with a matching outputs_hash, and returns a ResumePlan that the runner consumes verbatim. Pipeline blueprint changes or config drift trigger a full restart and cascade-invalidate the journal rows so a stale completed-prefix never confuses a future resume.

Classes

ResumePlan(run_id, restart_index, ...)	Decision returned by ResumePlanner.compute().
ResumePlanner(journal, workspace)	Decide where the pipeline must restart after a crash or interruption.
StepInvalidation(step_order, step_name, reason)	Trace of a single step the planner marked aborted.