fix(DateInput): keep cleared calendars empty

[korkt-kim]

User impact

After selecting a date through DateInput's calendar, clearing the field and reopening the calendar still shows the previous date as selected. This fix keeps the empty field and Calendar selection consistent.

Fixes #6934

Scope after rebase

Rebased onto main at 2514d86d02ffa32b1c17b32523e1149eb86334fe and removed the portions already covered by:

• [Component Audit] DateRangeInput: keep range state truthful #7297: Calendar's controlled-empty support and DateRangeInput's clear behavior.
• [Component Audit] DateTimeInput: keep cleared calendars empty #7298: DateTimeInput's clear behavior.

The remaining change is the DateInput composition and the original locale-consistency example, with regression evidence and a scoped patch changeset. Calendar, DateRangeInput, and DateTimeInput have no diff against the rebase target.

Problem and solution fit

Calendar treats undefined as uncontrolled. Selecting a date updates its internal selection; passing undefined after DateInput clears lets that old selection become visible again.

DateInput now passes optimisticValue ?? null to Calendar, following the same controlled-empty boundary used by DateTimeInput in #7298. DateInput's public value type, onChange(undefined) clear callback, optimistic change pipeline, and picker presentation stay unchanged.

Expected behavior and authority

• DateInput's consumer documentation describes clear as resetting the selected value.
• Calendar's existing consumer documentation defines null as controlled-empty and omitted/undefined values as uncontrolled.
• Once the parent accepts clear, the reopened calendar should have no selected gridcells. Visible month, today's marker, and keyboard focus remain separate from selection.
• No public API or type widening remains in this PR.

Smallest restoration

• Normalize DateInput's empty value to null only at its Calendar boundary.
• Keep the locale-consistency example's Calendar controlled when its shared date is empty.
• Add a select → clear → reopen unit regression and a ClearedCalendarSelection Storybook play function, following [Component Audit] DateTimeInput: keep cleared calendars empty #7298's evidence pattern.
• Narrow the changeset to DateInput; retain contributor attribution.

Evidence

• Before: without the DateInput boundary normalization, the new regression fails with one selected gridcell instead of zero.
• After: the test verifies that a calendar selection exists before clear, that clear empties the input, and that reopening leaves zero selected gridcells.
• Browser endpoint: Core/DateInput/ClearedCalendarSelection selects April 15, clears, reopens, and asserts an empty field with no selected day. Its final open-calendar frame is tagged visual-baseline.
• Representative unchanged paths: related suites cover ordinary selection, constraints, callbacks, native and touch presentations, and the already-merged Calendar/DateRangeInput/DateTimeInput clear behavior.
• No CSS, geometry, token, role, or DOM-anatomy change. Only the stale selected-day state is removed. This is not a new cross-browser or assistive-technology audit.

Scope

• One remaining stale-selection defect is restored.
• Already-merged sibling and Calendar changes are removed.
• No new public API, visual representation, default, or interaction model is introduced.
• Public text and artifacts contain no internal operational context.

Testing

• pnpm build — passed.
• pnpm -F @astryxdesign/core typecheck and pnpm -F @astryxdesign/storybook typecheck — passed.
• Commit hooks (staged-file ESLint/Prettier and pnpm check:repo) — passed; the user-reviewed patch is unchanged by the hooks.
• pnpm lint:strict --ignore-pattern 'assets/**' — passed. The plain command picked up a local, ignored generated bundle under assets/; that directory contains no tracked files. No repository lint rules or configuration were changed.
• pnpm test --maxWorkers=4 — 22,668 passed, 2 failed, 2 expected failures, 40 skipped. All DateInput and related date-component suites pass. The only failures are the two taxonomy assertions in apps/storybook/.storybook/story-tree.test.ts, which read a pre-existing local Storybook index. Both failures were independently reproduced with the exact main version of that test and the same index; the test itself is unchanged. They are not reported as passing and are outside this fix.
• Headless Chromium, Core/DateInput/ClearedCalendarSelection — passed: empty field, reopened calendar, zero selected gridcells, no page errors.
• git diff --check — passed.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated
astryx	Ready	Preview	Oct 11, 2026 2:05pm UTC

[astracat-bot]

AI review status for this pull request.

Review status	Updated
🟢 Reviewed (for maintainers only)	Oct 11, 2026, 2:30 PM UTC

[astracat-bot]

Thanks for the focused fix and the clear before/after tests — clearing a date input after a calendar pick no longer leaves the old date selected on reopen.

Approval needs one maintainer decision first: this introduces value={null} on Calendar as the controlled “no selection” value (with undefined staying uncontrolled), and there is currently no governing Calendar spec for that distinction. A maintainer needs to provide or identify the spec that settles Calendar controlled-empty in both single and range modes, including how it interacts with defaultValue and that onChange never emits null.

No other changes are requested: the select → clear → reopen behavior is covered for all three inputs, and existing uncontrolled use is preserved.

[Automated review]

[github-actions]

PR Analysis Report

📚 Storybook Preview

View Storybook for this PR

🧪 Sandbox Preview

View Sandbox for this PR

Modified Components

DateInput _{(@astryxdesign/core)} · View in Storybook

Metric	Before	After	Delta
Bundle Size (ESM)	N/A	N/A	N/A
Lines of Code	N/A	3374	-
Complexity	N/A	Very High (362)	-

Bundle Size Summary

Package	Size (ESM)	Size (CJS)	Gzipped
@astryxdesign/core	N/A	4.9KB	1.2KB

Accessibility Audit

Status: No accessibility violations detected.

Visual Regression

26 of 458 shot(s) changed.

A change here is a question, not a failure: check whether the after is the
picture you intended. Record that review on the PR. Baseline maintenance is an
explicit dispatch of CI; this report never rewrites the baseline or adds a release gate.

component	story	theme	mode	pixels
Selector	Indicator Space Bottom Sheet Narrow End RTL	neutral	light	754,077
Selector	Indicator Space Bottom Sheet Wide Start	neutral	light	740,188
Chat	Full AI Chat	probe	light	155,733
Chat	Full AI Chat	probe	dark	155,005
Chat	Default	neutral	dark	33,205
Chat	Default	neutral	light	32,876
Chat	Empty State	probe	light	23,531
Chat	Empty State	probe	dark	22,622
ChatAutoScroll	Scroll Behavior Comparison	neutral	dark	15,378
ChatAutoScroll	Scroll Behavior Comparison	neutral	light	14,107
Chat	With Attachments	probe	light	13,229
Chat	With Attachments	probe	dark	11,065
TableRowExpansion	Detail Panel	neutral	dark	5,129
TableRowExpansion	Detail Panel	neutral	light	4,050
SideNav	Resizable in AppShell	probe	dark	3,208
SideNav	Resizable in AppShell	probe	light	3,208
TableGroupedRows	Default	neutral	light	2,133
TableGroupedRows	Default	neutral	dark	2,116
Markdown	Default	probe	dark	897
Markdown	Default	probe	light	897

and 6 more.

---

_{Generated by PR Enrichment workflow | Storybook | Sandbox | View full report}

[korkt-kim]

Thanks @cixzhang! After seeing yesterday's fixes in #7297 and #7298, I rebased this PR onto the latest main and removed the overlapping Calendar, DateRangeInput, and DateTimeInput changes.

This now covers only the remaining DateInput controlled-empty boundary and the locale-consistency example, with regression tests. The DateInput fix follows the same ?? null pattern as #7298, and includes a select → clear → reopen Storybook regression.

[astracat-bot]

Thanks for narrowing this to the remaining DateInput piece: keeping the popover Calendar controlled while the field is empty means a cleared DateInput no longer reopens with the old date selected, matching the pattern DateTimeInput already uses on main. The new select → clear → reopen regression test covers exactly that outcome, so this automated review requests no changes; the pull request still needs the usual maintainer review before merging.

[Automated review]