Fixture Contribution Guide

Reduced fixtures help BDS support real cycler exports without asking users to share full private experiments. A good fixture is the smallest file that still reproduces detection, parsing, unit conversion, current-sign, or validation behavior.

What To Keep

• Keep the original header rows, units, delimiter, worksheet name, and encoding whenever possible.
• Keep the columns needed to reproduce the problem, especially time, voltage, current, status/step type, cycle, step, capacity, energy, temperature, and any vendor-specific metadata rows that affect parsing.
• Keep enough rows to show the behavior. Five to twenty rows is often enough; use more only when the bug depends on a repeated cycle, a step transition, or a sampling gap.
• If the current sign is important, keep at least one charge row and one discharge row, plus the source status column if it exists.

What To Remove

• Remove or replace cell IDs, customer names, project names, operator names, sample names, serial numbers, barcodes, file paths, and comments that reveal confidential work.
• Remove long procedure descriptions unless the parser needs those rows to find the data table.
• Remove most measurement rows after confirming the reduced file still fails or still demonstrates the requested mapping.
• Replace proprietary values with simple representative values only when doing so does not change units, signs, headers, step transitions, or parser behavior.

How To Submit

1. Run bds explain <file> --text on the reduced file.
2. Confirm the reduced file still shows the same detection, conversion, or validation behavior as the original.
3. Open a cycler import issue and paste the bds explain output.
4. Attach the reduced file if policy allows it.
5. State clearly whether the file may be added to public regression tests under tests/fixtures.

If the file cannot be shared publicly, include the header row, cycler model, software/export version, sheet name, current sign convention, and a short description of the expected columns.