Standard UI Components

The delay keeps fast responses from flashing a spinner that reads as a glitch; the minimum display time stops a spinner that did appear from vanishing a split second later. Both remove flicker; neither slows fast responses.

An indicator is determinate only when the product can measure completed work against a known total; otherwise it is indeterminate. Faked percentages stall, jump or run backwards, and teach users to distrust every progress bar.

The parent shows checked when all six are checked, unchecked when none are, and mixed, drawn as a dash, when only some are. Activating it checks all six, or unchecks all when all were checked; mixed is derived, not picked.

Use a filterable combobox, not a scrolling list: typing narrows the 2,000 options, the popup stays open for more picks, selections show as removable chips with an overflow count, and grouped or virtualised options still report their position and the total.

With automatic activation, moving focus to a tab shows its panel; with manual activation, focus moves freely and Space or Enter shows the panel. Choose manual when panels cannot appear without noticeable latency.

The tab list is one stop in the Tab sequence: Tab lands on the active tab, arrow keys move between tabs and wrap at the ends, and the next Tab press leaves the list, usually into the panel.

A multi-line text area signals that sentences are expected, wraps text and lets Enter add a line. Its counter states the limit up front, shows what remains, flags and blocks over-limit text instead of cutting it, and announces only at thresholds.

Try the preferred side, flip to the opposite side if that overflows, then shift along the edge to stay inside a safe margin with the arrow still on the trigger; if nothing fits, cap the size and scroll inside.

Dots that differ only in hue fail people with color-vision deficiencies (WCAG 1.4.1), low-vision users when the dot lacks 3:1 contrast against its surroundings (1.4.11), and anyone relying on a text alternative. Fix with distinct shapes, a separating ring, and text.

Check that each button has an accessible name describing its action, that targets meet 2.5.8's 24-by-24 minimum or its spacing exception, that icons reach 3:1 non-text contrast, and that toggle buttons keep a stable label while exposing state.

One message hides three different causes, and the error case is false. Design each state separately: explain first-use and offer the next step, show and relax filters for no results, and show an error with retry when data failed.

Users click through confirmations that look the same every time. Name the prescription and the consequence, label buttons with verbs, focus the safe choice, and where possible replace the confirmation with an undo so the action is reversible.

Stacked modals are allowed but fragile: each layer must block those beneath, Escape closes only the top one, and focus returns to its trigger in the dialog below. A multi-step task needing history or its own address belongs on a page.

Route every toast through one region with a small visible limit, queue the rest and collapse duplicates, start each timer only when shown, and escalate the failed payment out of toasts into a persistent banner.

Measure where guests leave, give each step one clear topic, show named progress in text, keep every answer on back navigation, never ask for the same data twice, and add a review step before the card is charged.

Generate from a schema when many similar forms must follow one pattern, ideally across platforms; the schema must then carry labels, help, grouping and order, not just types. Hand-compose the few high-stakes flows that need bespoke layout and copy.

Name the indicator, announce the wait and occasional progress without moving focus, announce completion or failure explicitly, and mark the updating region busy. WCAG 2.2 SC 4.1.3 Status Messages treats waiting and progress as status messages.

The placeholder and the real content occupy different space, so arriving data pushes rows under the user's finger. Fix it by building skeletons from the real layout, reserving space for late media, and never inserting content above what the user is reading.

In a radio group, arrow keys move focus and check the option in one step, so reaching the third rate fires two reloads that throw focus away. Update the price in place without moving focus, or apply the choice on an explicit action.

It uses list autocomplete with automatic selection, so a partial name becomes the highlighted first match when focus leaves. For a costly pick from a fixed set, use manual selection: typing only filters, someone must choose explicitly, and unmatched text is flagged, not replaced.

Carousels hide most of their content behind time, users rarely see or act on later slides, and rotation interrupts reading. Propose one prioritised feature plus a visible list or grid of the rest, and measure it.

Strict masks reject or garble pasted and autofilled values, fight caret movement and deletion, assume one format that real values break, and can be read aloud as symbols. Specify lenient input, silent normalisation, and a format example in helper text.

Keyboard, switch and screen-reader users cannot move focus into a tooltip, and touch users cannot hover, so most people never reach the link. Replace it with inline text and a link, or a popover opened by an explicit tap.