Front-end development: responsibilities and practical exercise
Front-end development turns content and design rules into usable browser behaviour. It connects HTML structure, styles, interactions and received data. Review the work through a complete journey, including keyboard use, late responses and states where results are not yet available.
These sheets describe activities and teaching exercises. They do not constitute job offers, certifying courses or recruitment promises.
Implement browser interfaces: HTML structure, presentation and interactions connecting content and services with visitor actions, performance and accessibility.
Typical activities
Build semantic components for different screens.
Handle loading, network errors and empty states without losing input.
Test keyboard use, browsers, long content and API interactions.
Responsibilities to agree with the team
Browser validation supports usability but cannot secure data alone. Server checks remain essential; agree API contracts and errors with back-end developers.
A first piece of work to demonstrate
Build a keyboard-usable form and demonstrate slow-response and simulated-server-error states.
This is an editorial description of common activities. Job titles and scope vary across teams; it is not a certification framework.
Create a catalog filter that can be used on the keyboard and shareable by a URL.
Work on a copy or demo environment. Define the data used and the expected result before you begin. Keep your decisions and the difficulties encountered in order to be able to explain the process.
Define states and HTML structure before interactive behavior.
Predict absence of javascript, loading and zero results as needed.
Check keyboard, focus, long texts and small screens.
Document dependencies and regression controls.
Deliverable and checks
Deliver the catalogue filter with data fixtures and URLs reproducing selections. Describe loading, no results, errors and reset. Include checks for focus and restoring the selection after reloading the page.
Open a filtered URL directly, change the selection by keyboard and reset it. The result count, visible list and address should agree. If responses arrive out of order, only the still-active selection should determine the display.
Mouse operation on a single screen is not enough. The interface must announce its changes and maintain an understandable state.
Collaboration to explore: Design UI.
Fictional deliverable excerpt : The email field keeps its value after an error and displays an explanation associated with the field. Confirmation appears when processing allows it.
What passes between the roles : Both sides need a shared understanding of fields, accepted formats and possible responses. Checking the journey means examining failure, success and actual enquiry handling beyond the appearance of the button.
Fictional teaching material: this excerpt illustrates part of a deliverable. It is not a client reference, a complete specification or evidence of professional qualification.
Sample excerpt
<label for="example-search">Filter resources</label>
<input id="example-search" type="search"
aria-describedby="example-help">
<p id="example-help">Enter a word from the title.</p>
Boundary case: two filters appear on the same page.
Expected: each label and help text target the corresponding field, using distinct identifiers.
What the author should be able to explain
The for value exactly matches the field identifier.
The label remains visible after typing begins.
Help supplements the label rather than relying only on placeholder text that disappears.
Does this fragment implement a working search?
It demonstrates the association between field, label and help. Filtering, empty results and result-count announcements need additional behaviour, illustrated in the demonstration. Duplicate the fragment in a test copy, renaming id, for and aria-describedby together. Click each label and inspect the accessible name and description. Changing only the visible wording does not resolve duplicate identifiers.
Another case to examine
Add an error without losing field help
Fictional case: a field has permanent help, then receives an error after validation. The error text must remain associated with the relevant field.
In the test copy, give the error a distinct identifier and reference both help and error in aria-describedby. Inspect the accessible description rather than only their visual proximity.
After correction, remove the reference to the vanished error while retaining the help reference. Check that remaining referenced identifiers exist and that input has not been cleared.
These three fictional entries show a result, an empty state and a return to the full list. Try “mobile”, then an absent word. No network request is made.
3 of 3 examples shown.
Images and alternativesPrepare an image and its description.
Mobile checksCheck a journey on a phone.
Writing a service pageClarify an offer and its scope.
For review: use the keyboard, test an accented word and check the count after clearing. This local list demonstrates neither server loading nor network failure. Input is not saved or sent.
Assess the quality of the work
For the exercise, a form loads choices from a remote service before submission. Examine keyboard controls, then distinguish waiting, an HTTP error response and network failure. Receiving a Fetch response does not establish business success; visible behavior should remain consistent with the status and data actually received.
Scroll the table horizontally to read every column. With a keyboard, focus the table area and use the arrow keys.
Assess the quality of the work
Skill
Evidence to examine
Failure to spot
Controls and names
Use elements appropriate to each action with understandable names and coherent focus order. Associate help and errors with the correct field.
An element works by click but not keyboard, or two buttons announce the same name despite performing different actions.
Error and recovery
Prepare a valid empty response, an HTTP error response and a network disconnection. Define the message, retained choices and retry behavior for each.
Every received response displays success, or retry reuses an invalid selection without informing the visitor.
Maintenance
The change is localised and dependencies and checks are known.
Fixing the component breaks other occurrences.
Late network response
Handle loading, failure, retry and obsolete responses.
An earlier search cannot replace the current results when responses arrive out of order.
A focused next step
In a test environment, trigger an empty response, HTTP error and slow response in turn. Replay recovery using the keyboard and check the visible result rather than only the console. Document recovery conditions; if the previous state is unusable, explain why it must be selected again.
Language and reading direction are separate information. Declaring ar alone does not establish the base direction; dir="rtl" does. Logical CSS properties adapt spacing without hard-coding left and right throughout the layout.
When one URL produces different responses according to Accept-Language, Vary identifies that cache selection criterion. Keep the value consistent on 200 and 304 responses. Separate language URLs do not require this header merely because the website is multilingual.
An authentication step requiring recall or transcription needs assistance or an alternative route, subject to the criterion’s exceptions. Test password managers and code pasting; an audio alternative requiring transcription does not by itself remove the cognitive burden.
Dates indicate when references were consulted. These facts are neither a site audit nor an overall compliance statement.
A review case to reproduce
Old search results replace new ones
Situation to examine
Search requests return in a different order from submission. The interface displays results for a term already replaced.
Proposed check
Simulate slow responses in a test environment. Ensure displayed data corresponds to the current query. Check waiting, cancellation, no results and focus before choosing an implementation.
Evidence to retain
A reproducible case records request order and final state. Displayed query and results belong to the same search.
Fictional finding example
Request Q1: “Paris”. Request Q2: “Lyon”. Q2 returns first with three results; Q1 then returns with eight.
Interpret this outcome
The final screen should still show “Lyon” and its three results. Testing response order does not establish handling of every network failure.
Three complementary checks
Fictional cases: adapt these expectations to your test contract and retain the limits of the check.
1Identified request
2Checked response
3Consistent display
Scroll the table horizontally if needed; with a keyboard, focus it and use the arrow keys.
Scenario variations and expected evidence
Variation to test
Expected result in this case
Evidence to hand over
Q1 arrives after Q2
Keep the current search Q2
Send order, response order and final state
An old request is cancelled
Do not replace the current result with its error
Cancellation reason and retained display
HTTP 404 is received
Handle the status; avoid a false successful empty result
Status, response and visible message
Decision to hand over : Link every displayed result to its request; receiving a response does not automatically mean success.
Extended exercise: Integrate a filterable list with maps with titles, photos and links. HTML must remain understandable before JavaScript runs. Test a very long title, no response, a slow connection and a language that is written from right to left. Use images adapted to their displayed size and declared dimensions. Check the focus order after an action: Updating the list should not lose the context of the searching person.
What to deliver with your work
Structured HTML and demo dataset
Styles for multiple widths and reading directions
Filter script with report management
List of checks carried out in the browser
Acceptance criteria
These criteria are a working basis to adapt to the context. For each, note the action taken, the expected result, the observed result and the information remaining to be confirmed.
Essential content and its links remain accessible without JavaScript.
The filtering works on the keyboard and announces its result in an understandable way.
The images have relevant alt text and consistent dimensions.
An empty state provides an explanation and resets the search.
Document dependencies, useful commands and structure decisions. Keep the test data that reproduces long, empty and multilingual cases. When handing over, explain how to add a card and what checks to start again. Proper rendering in a single browser is not enough to conclude on all the conditions of use; Indicate the actual controlled environments.
Choose interactivity according to the task. A readable page and a rich application require different dependencies, states and loading conditions.
Progressively enhanced HTML
When to choose it
Suitable for content, links and forms that work with a server response. Add interactions where they support an identified action.
What to plan for
Check use without scripts, errors and slow loading. Messages should remain associated with fields, and keyboard users should be able to complete the task.
Interface with browser-side state
When to choose it
Useful for frequent interactions such as filtering, comparison or editing. Define the displayed data and states precisely.
What to plan for
Plan history, interruptions and synchronisation. Do not announce success before an important operation is actually confirmed.
A risk to anticipate
Rendering at one width does not validate components. Long content, zoom, translations and missing data change layouts.
How to decide
Test the same task with a keyboard, on a small screen and with a network interruption. Provide a way to reproduce observed errors.
Compare components, keyboard behaviour, adaptation and service exchanges. A static page differs from an interface handling network failures.
Recurring operation
Plan dependency updates, regression checks and compatibility monitoring. Separate upkeep of existing components from development of new behaviour.
Exit and handover
Request sources, build commands and interface contracts. Check that another team can produce the served version; minified files alone do not replace these assets.
How can an interface be tested before the API is ready?
Use synthetic responses matching the contract: success, empty list, error, delay and long values. Identify them as simulations. Once connected, repeat the cases against the server: interface testing proves neither authorisation nor business recording.
What should be checked when search responses arrive out of order?
Simulate a slow first request followed by a faster second one. Displayed results should match the current input, with understandable pending feedback. Keep the request timeline and result identifiers; the final appearance alone may hide an older response briefly shown.
Does a framework guarantee an accessible interface?
It may provide useful conventions, but structure, labels and behaviour depend on implementation. Check the final workflow with its data, error states and screen adaptations. Choosing a tool does not replace these checks.