frontend-patterns/research/candidates/FocusReturn.md

title	status	maturity	families	scales	admission
Focus Return	draft	hypothesis	quality affordance	L6	research-candidate-only

Focus Return

Context: a modal task temporarily interrupts another keyboard workflow. Closing it can leave users without a meaningful continuation point. Continuity competes with changing document structure and the next task's needs.

Proposed solution: retain the invocation context and choose a meaningful focus location inside the modal. On closing, return focus to the invoker when appropriate; if it disappeared or completion changes the next step, choose a logical surviving target. Define ownership for nested dialogs so an obsolete close callback cannot move focus out of the current task.

Result: interruption has an explicit return path, but dynamic targets need lifecycle handling. Do not steal focus on unrelated background updates or apply modal focus containment to ordinary page regions. This narrow candidate does not claim complete accessibility conformance; labeling, inert background and keyboard behavior remain part of the containing dialog contract.

Evidence: B06 locates APG's modal focus guidance and return exceptions. Generalizing to route transitions requires separate evidence. Alternative: inline interaction when modal interruption adds no value. Defer merging this into a broader accessible interaction pattern until contrasting cases support that scope.

Proposed relation: can support Explicit Reconciliation when comparison uses a modal; reconciliation does not require a modal.

Next test: close nested dialogs, delete the invoking row, and complete a task that creates the next target; verify keyboard and assistive-technology behavior.

Evidence locators: breadth review. All relations are research proposals; no canonical identity or admitted graph edges. The next test is planned, not executed. Known uses of the complete proposed contract have not been verified in this investigation.

Executed partial evidence — 2026-09-26

Boundary experiment X05–X08 runs Radix Dialog 1.1.23. Ordinary return succeeds; removal of the trigger leaves BODY focused unless the application supplies a surviving target. Nested dismissal is timing-sensitive: immediate Escape failed the parent-return obligation in the final recorded immediate run, while a 100 ms settling variant passed. Neither focus arrival nor dialog visibility alone proves dismissal ownership has stabilized. The delay is an experimental variable, not a recommended fix. Root cause and accessibility conformance remain unverified; retain hypothesis maturity and test lifecycle ownership before a stronger claim.

Nested ownership isolation — 2026-09-26

Follow-up matrix captures an interval where the inner layer is registered and focused but the outer Escape listener still handles input. This explains the recorded wrong-owner dismissal in the instrumented fixture. Development/production and Chromium/Firefox comparisons retain all 48 trial results. A current-owner guard can lose the keypress; private listener-readiness waiting passes its trials but is not a product solution. Test ownership of dismissal as well as return focus, including activation/teardown input. No arbitrary-delay fix, production repair or full-pattern promotion is established.

Experimental routing evidence — 2026-09-26

Routing comparison tests layout registration, a single document Escape dispatcher and event-time owner selection. The experimental router meets 24/24 scenario assertions across two browsers/builds; its unmodified control meets 20/24. It preserves cancellation and removes ownership on teardown without a delay or private-readiness wait. Return to a replacement target still requires application policy. This supports separating input ownership from focus restoration, not a general production-ready fix: animated exits, competing owners, rapid input bursts and assistive technology remain untested. Sixteen hypotheses; no new pattern or maturity promotion.

Boundary with route continuation — 2026-09-27

Navigation focus contrast NF02–NF06 provides a non-modal counter-context: a successful route may continue at a new heading, while failed navigation keeps the invoker and background work leaves focus alone. A held response must respect focus moved to a surviving search field. Raw htmx in this fixture restores a same-ID link; that behavior alone does not choose the next task. Preserve this candidate's modal scope instead of treating every route change as a return-to-invoker problem. No general route-focus pattern or AT validation is admitted; delayed history restoration and competing navigation remain open.

Completion ownership contrast — 2026-09-27

NO01–NO06 strengthen the route/modal scope boundary. A still-current history result can restore content without taking focus from a newer search task; an obsolete navigation result must not supply a continuation at all. The earlier single-response route policy is insufficient for competing completions. This contrast does not expand the modal candidate into a universal routing pattern or prove AT behavior.

Independent non-modal source contrast — 2026-09-27

Zulip ZN05–ZN07 locates search entry/exit focus policy and selection-aware editing restoration. These are task-specific continuations, not evidence of returning from a modal to its invoker. The inspected editing restoration schedules focus work after a selected-message check; this does not prove ownership is valid when the timer runs. Retain the modal scope and mark AT/runtime validation and auxiliary continuation behavior unknown.