Run history and replay
Run history lets you return to the output, variables, and figures from an earlier execution. Use it to compare two results, inspect a failed run, or continue an analysis after reopening the project.
Enable Persist runs in Settings when runs should remain available after the current session.
Open an earlier run
Use the run selector in the Runtime area to choose a completed, failed, or cancelled run. Output, Variables, and Plots update to show the information saved for that run.
Each run records its source file or notebook, execution target, status, and time. Depending on persistence settings and the values produced, it can also include:
- command output and diagnostics;
- a workspace summary and inspectable variables;
- interactive figures and exported images;
- trace events and other saved artifacts.
Saving large variables and figures can continue briefly after computation finishes. The saving indicator disappears when the run record is ready.
Compare results
When a result changes unexpectedly, compare runs in this order:
- source file or project version;
- inputs and configuration;
- execution target and Runtime version;
- command output and diagnostics;
- variables and figures.
This order helps distinguish a source or input change from a presentation difference. Use a notebook when the comparison needs a written explanation beside the code and figures.
Replay saved values
Selecting an earlier run can restore supported variables and figures for inspection while leaving the active execution workspace alone. This is useful for exploring an earlier result in Variables, Plots, Agent, or the Command Window.
Replay depends on the data saved with the run. Numeric arrays, structures, and supported figures can be restored. External files, network connections, operating-system handles, and other live resources may require a fresh run.
An unavailable value means its data was not saved, has been removed by retention policy, or cannot be restored in the current environment. Rerun the source to recreate it.
Reproduce a result
Replay helps you inspect a previous result. A clean run checks whether the current project can produce it again.
For a reproducible result, preserve the source revision, runmat.toml, package lock state, input data, Runtime version, and execution target. Run the normal project entrypoint from a fresh workspace before treating an earlier output as current.
Run history and file history
Run history records executions and their results. File history records changes to project files. Cloud projects also support project snapshots that capture a set of file versions.
Use file history to compare or restore source. Use run history to inspect what the source produced. See Cloud collaboration and History and synchronization for cloud project history.