Ask what a second action means
A small automation needs a definition for repeated use. If someone activates the same control twice, should the exercise restart, ignore the duplicate or explain that work is already in progress? Choose the behaviour deliberately. A result should not depend on an accidental second event handler or the learner guessing whether the first click registered.
Use a fictional task with no external effects while exploring the behaviour. Do not demonstrate duplicate handling by sending real messages, creating real bookings or making purchases. A browser-local sample can reveal the same design question without involving another service or account. Keep the expected state visible in the exercise notes.
Describe the states in plain language
Write a small state list such as ready, reviewing, finished and stopped, then define what each allows. These are illustrative labels, not a required software framework. The important point is that the interface and program agree about what is happening. An enabled control should not promise an action the current state cannot support.
For a fictional Bulawayo coding workshop, a learner might rehearse the sequence with cards before implementation. Another participant activates the proposed actions while the first follows the state list. Where the sequence becomes ambiguous, revise the explanation. This is an original design exercise, not evidence that a production system has passed a resilience test.
Keep cancellation honest
MDN documents AbortController for supported abortable operations. That does not mean a stop button can undo every completed action or cancel an unrelated process. Explain exactly what a control stops in the actual implementation. If a future project involves external work, distinguish cancelling a local wait from reversing a request already accepted elsewhere.
Do not display cancelled as a convenient substitute for unknown. If the program cannot establish the outcome, preserve that uncertainty for review. A learning tool should not claim that no information left the device merely because its visible panel closed. Privacy and external-effect claims require inspection of the actual code and services.
Test the uncomfortable sequence
Try the ordinary start and finish, a repeat activation, a stop before completion and a fresh start afterwards using the fictional sample. Record the expected and observed states. Keep the test bounded; there is no need to hammer an endpoint or simulate load against a service you do not control.
Use the findings to simplify the task before adding more automation. A reliable small exercise teaches state and user control more clearly than an elaborate demo with unexplained outcomes. It does not prove that a bot is ready for regulated work, confidential records or unattended actions with real-world consequences, and it cannot replace the appropriate organisational review.
Sketch the screen immediately after stopping. The reader should be able to identify the preserved draft and the action required to begin a fresh attempt without guessing.
Sources & further reading
Use these references to check the underlying guidance and any current local requirements. Practical worksheets and illustrative examples in this guide are original educational material.
This guide is general information. Source notes are not a claim of independent professional review.