Skills practice questions.

A student writes 'The program will be nice to use' as a success criterion for the computational solution. Which replacement can actually be used in the evaluation?

• The program will look modern and behave in a professional way.
• The program will store the data in arrays and write them to a file.
• A club official can record a new member in fewer than four screen actions.
• The program will be tested by the developer until it no longer produces errors.

Answer: A club official can record a new member in fewer than four screen actions.

A success criterion has to be checked by somebody else against the finished product, so it names an observable outcome with a measurable limit. Criteria about appearance, about the code itself or about the development process cannot be tested by the client.

A student cannot find an external client for the computational solution. What is the accepted way to continue?

• Work with an adviser, who takes the client's place in the consultations.
• Invent a client and write up the consultation notes from imagination.
• Choose a problem that needs no user at all, such as a puzzle solved for fun.
• Ask the teacher to approve the finished product once the code is working.

Answer: Work with an adviser, who takes the client's place in the consultations.

The solution must answer a real need that somebody else expresses, so where no client is available an adviser gives the same feedback on scope, criteria and the finished product. A fictional client leaves the success criteria unchecked and makes the evaluation worthless.

What is the main purpose of the Record of Tasks in the computational solution?

• To show the client which features were finally delivered.
• To document, in date order, what was planned and achieved at each stage.
• To list the sources of any code that was adapted for the product.
• To record the outcome of every test together with the exact data that was used.

Answer: To document, in date order, what was planned and achieved at each stage.

The Record of Tasks is a running log kept while the work happens, so it evidences planning and decision-making over time. Delivered features belong in the evaluation, sources in the development section and test results in the testing evidence.

A field stores the number of goals scored in a match, a whole number from 0 to 20. Which set of test data covers a normal case, a boundary case and an erroneous case?

• 0, 20, 21
• 1, 2, 3
• 21, 25, 'nine'
• 7, 20, 'nine'

Answer: 7, 20, 'nine'

7 lies inside the range, 20 sits exactly on the upper limit and 'nine' is not a whole number at all, so the three values test acceptance, the limit and rejection. A set built only from middle values never shows whether the validation works.

How should the evaluation section of the computational solution be organised?

• As a paragraph describing how much the student learned from the whole project.
• As a judgement against each success criterion, supported by client feedback.
• As a list of the programming techniques used in the finished product.
• As a summary of the errors found and corrected during development.

Answer: As a judgement against each success criterion, supported by client feedback.

The criteria agreed at the start are the yardstick, so each one is judged separately and backed by what the client says about the product. Techniques belong in the development section, and corrected errors are testing evidence rather than evaluation.

A student writes the program first and draws the flow diagram afterwards so that it matches the finished code. Which part of the work is weakened most?

• Planning, because the Record of Tasks then holds fewer entries.
• Functionality, because the video can no longer show the product working.
• Design, because the diagram no longer shows decisions taken before coding.
• Evaluation, because the criteria can then no longer be tested by the client.

Answer: Design, because the diagram no longer shows decisions taken before coding.

The design has to document the thinking that guided development, so a diagram reverse-engineered from finished code evidences no decision at all. The other stages still produce their usual evidence.