T Level Digital: Development and Delivery

Four of these ten questions are about testing, and they are deliberately built as a chain: each one names something the previous kind of testing cannot tell you. A function that passes every unit test still says nothing about what happens when the rest of the system calls it. A change that integrates cleanly can still have broken something elsewhere, which is what regression testing exists to find — and the point worth holding onto there is that the fix may be entirely correct; regression testing is about the side effects of a correct change. And software that passes every test the developers can write can still fail the only test that matters, which is whether it fits the way the people using it actually work. Around that chain sit the practices that make delivery possible: what the so-that clause of a user story is really for, what acceptance criteria settle in advance, why dated copies of a folder are not version control, what a daily stand-up is for and how it goes wrong when it turns into a status report, why a start-up might ship one feature, what four copies of the same pricing logic cost later, and what a staging environment catches. Zestly is an independent study tool. It is not affiliated with any exam board and it is not an exam centre.

• Read a user story and say what its so-that clause is for
• State what acceptance criteria settle, and why they are written before the work starts
• Say what unit testing establishes and what it leaves open
• Explain regression testing as a check on the side effects of a change rather than on the change itself
• Say what user acceptance testing decides that developer testing cannot
• Give the real difference between version control and dated copies of a folder
• Explain the purpose of a daily stand-up and how it fails when it becomes a status report
• Justify a minimum viable product, recognise duplicated logic as a future cost, and say what a staging environment catches

A developer has unit-tested a VAT function and it returns the correct figure for every input she tried. What has that NOT yet shown? — Whether the function works correctly when the rest of the system calls it, with real data flowing in from other parts.