Test a dirty read is prevented

Simulate a reader and an uncommitted writer and prove the reader's view never contains the writer's rolled-back prices.

1. 1Reader: capture committed camera prices with CREATE TEMP TABLE reader_snapshot AS SELECT id, price FROM products WHERE category = 'cameras'.
2. 2Writer: BEGIN; halve every camera price (price = price * 0.5).
3. 3Writer aborts: ROLLBACK.
4. 4Verify: SELECT COUNT(*) AS dirty_rows FROM products p JOIN reader_snapshot s ON s.id = p.id WHERE p.price <> s.price returns 0.

• temp.reader_snapshot holds the committed camera prices
• The last result set shows 0 dirty rows
• The script uses ROLLBACK

• Every run starts from a freshly seeded database, so your script must include its own setup statements.
• The workbench has a single connection, so concurrent sessions are simulated step by step inside one session (snapshots in TEMP tables, SAVEPOINTs as the other transaction).
• Only the last result set of your script is compared, so finish with the verification query.