Repair a quantity update that loses zero

An inventory API accepts partial updates to a product. A missing quantity means keep the stored number, but an explicit zero means no stock remains. The supplied small example in the linked lab uses a truthiness fallback and therefore loses zero.

This is a focused review exercise. You will write the input/output contract, explain the failing expression, propose the smallest repair and demonstrate the real runtime inputs. You are not being asked to build a complete inventory service.

Your starting input and finish line

Start from current = 7 and send {"quantity": 0}. The broken expression returns 7. The repaired boundary must return 0, retain 7 when the field is omitted, and reject the invalid cases you explicitly choose. Keep invalid input from changing stored state.

Use the quantity lab and its four-line starting function. Read the code first. Ask an assistant to explain the distinction between a missing property, null and zero, then compare that explanation with the language behavior. Hand over the patch and a small transcript of the agreed cases, including any outcome that remains unsupported.

Your first engineering conversation

“A customer saves quantity zero, but the old value stays. An AI supplied the patch and passing tests. What should happen, what assumption is wrong, and what evidence would you require before accepting a repair?”

For this exercise, current 7 plus supplied 0 must save 0; omitted quantity keeps 7; negative or null input is rejected. Agree on those examples first. Then trace the request, name the invariant, ask for a small implementation, inspect its diff, and make an intentionally broken version fail the same test.

Start with the change loop and working with AI. Every chapter now opens with a concrete review problem. Project indexes let you choose one independently readable brief, with follow-up diagrams and checks.

Learn from the diagrams

The original diagrams remain the visual foundation. Follow moving requests, shrinking budgets, and multiplying retry branches. Use the coding route for focused algorithm animations and the production casebook for AWS failure mechanisms.

Apply the concepts to production failures

Use the production casebook to review an AI-generated design against recent incident mechanisms. Ask for the invariant, failure injection, recovery behavior, and evidence before accepting a patch.