1.19Question 2

Prompt: Your team's test suite uses SELECT * throughout all database verification queries. A new migration adds a JSONB column to the users table, and suddenly 40% of your tests fail even though no business logic changed. What happened, and how would you redesign the verification approach?

What a strong answer should cover:

• Understanding that SELECT * returns all columns, so adding a new column changes the result shape
• Positional access (row[0], row[1]) breaks when column order changes
• The fix: use explicit column lists and DictCursor for named access
• Test resilience as a design principle

Example answer:

• The tests break because they use positional indexing (row[0], row[1]) against the results of SELECT *. When the new column is added, the positions shift for any columns that come after it in table order.
• I would refactor every verification query to list columns explicitly: SELECT id, name, email, role FROM users instead of SELECT *. Then I would switch from positional access to DictCursor so assertions read user["role"] == "viewer" instead of row[3] == "viewer".
• Going forward, I would establish a team convention: no SELECT * in test automation, mandatory DictCursor for readability, and a review checklist item for schema resilience.