When you attach a dataset to an experiment as historical data and open the mapping editor, you may see this warning on a row:
Cannot add a constant to a dataset that is used as output dataset for an experiment.
This article explains what this means and what you can do instead.
What Is an Output Dataset?
An output dataset is a dataset that was generated by an SDLabs experiment — it contains the results of runs submitted through the platform. When you browse and attach an existing dataset that came from a previous experiment, it is treated as an output dataset.
Output datasets are different from files you upload yourself: their columns are already tied to the parameters and measurements of the original experiment, and their structure is fixed.
Why Constants Are Blocked
A constant in the mapping editor creates a synthetic column — it assigns a fixed value to every row in the dataset for a given parameter or measurement. For output datasets, this is not yet supported.
Only column mapping is available: each parameter or measurement in your experiment must be connected to an existing column in the output dataset.
Live Link Between Experiments
When you attach an output dataset to a new experiment, the two experiments are linked. Any new data submitted to the original experiment is automatically included in the new experiment's historical data — the dataset is not a snapshot, it reflects the current state of the source experiment at all times.
If you want a fixed, static version of the data instead, download the output dataset as a CSV and upload it as a new file. Uploaded files are independent of their source experiment and will not pick up future changes.
What to Do Instead
Download the output dataset as a CSV, add a new column with your constant value for every row, and upload the modified file as a fresh dataset. Uploaded files are not treated as output datasets, so the constant option will be available when you configure the mapping.
For a full guide to the historical data mapping editor, see Attaching Historical Data to an Experiment.