# Edge-Test Checklist

Appendix C Kit 2, in checkable form. Run the set on a copy, never the real file. Anything unexpected in the last column is a bug found while it was cheap.

## Setup

- [ ] I made a copy of the file. The original stays sealed.
- [ ] I will run every row below on the copy only.
- [ ] I have the expected outcome written down before I run each test.

## The Six Tests

| # | Edge case | How to set it up | Expected outcome | Actual (fill in) |
| --- | --- | --- | --- | --- |
| 1 | Zero | Put 0 where a rate or revenue belongs. | A 0 result or an explicit flag, never a silent blank. | [ ] |
| 2 | Negative | Enter a credit or a refund. | Signs stay correct. No absolute-value surprise. | [ ] |
| 3 | Blank | Leave a required cell empty. | A visible error or a deliberate default, never a silent wrong number. | [ ] |
| 4 | Huge | Double the biggest value your data has ever seen. | Totals still add up. Nothing overflows into the next column. | [ ] |
| 5 | Duplicate | Paste the same row twice, same ID. | The total doubles on purpose, or the file warns you. No quiet collapse. | [ ] |
| 6 | Text in a number column | Type N/A into a numeric field. | A loud error or a flag. Not a quiet zero. | [ ] |

## After the Run

- [ ] Every unexpected result has a note about what it did and what it should have done.
- [ ] Each failure is classified: loud (easy to spot) or silent (dangerous).
- [ ] Failures got fixed on the copy, and the set was re-run.
- [ ] Nothing touched the real file until the whole set passed.

A sheet to practice on ships in the kit's `data/` folder as `commission-fake.csv`.