Change without losing the target

Your data workbench

Our model: kits.id ← loans.kit_id. One material can have many loans; names may repeat. We use fictional data only.

Predict before opening the reference. It shows verified output from the original example, not execution of your text. Run your changes with the kit on your computer.

-- Change without losing the target
BEGIN;
SELECT id, stock FROM kits WHERE id=1;
UPDATE kits SET stock=4 WHERE id=1;
SELECT id, stock FROM kits WHERE id=1;
ROLLBACK;
SELECT id, stock FROM kits WHERE id=1;

[
  {
    "columns": [
      "id",
      "stock"
    ],
    "rows": [
      [
        1,
        3
      ]
    ]
  },
  {
    "columns": [
      "id",
      "stock"
    ],
    "rows": [
      [
        1,
        4
      ]
    ]
  },
  {
    "columns": [
      "id",
      "stock"
    ],
    "rows": [
      [
        1,
        3
      ]
    ]
  }
]

Run the command in the extracted folder’s terminal. Use python3 on macOS/Ubuntu or py if that is your Windows launcher. Every run starts in memory; read README.md for saving files and recovery practice.

python lab.py run lesson.sql

The editor saves your .sql draft in this browser. Download it and replace lesson.sql in the kit; for the final project, availability.sql. The ZIP remains the original template. GitHub publishing sends only the confirmed file, without executing SQL or sending a database.