fidelity-ai-workspace/project-knowledge/02-work-items/pdiap-15836.md

type, project, status, ticket, title, systems, workstreams, people, related, updated, tags

type	project	status	ticket	title	systems	workstreams	people	related	updated	tags
work-item	fidelity	in-progress	PDIAP-15836	Modernize dismissal delegate lifecycle sequencing for pure SwiftUI environment	xflowsdk xflowviewmaker ftframeworks	xflow-swiftui-migration consumer-integration	jeff-dewitte	pdiap-14859 pdiap-12284 pdiap-15838	2026-05-14	work-item fidelity xflow swiftui

PDIAP-15836 - Modernize dismissal delegate lifecycle sequencing for pure SwiftUI environment

Status

• Moved to In Progress on May 7.
• David found a possible minimal dismissal fix path that would not require removing UIHostingController, and was validating it.
• Jeff directed David to do this dismissal/lifecycle work together with PDIAP-12284 in the same branch because both changes are disruptive enough to require consumer testing.
• May 11 simulator log review suggests the tested SwiftUI-host path now fires in the intended order: host-disappearance confirmation precedes fireEndActivityDelegates, delegate callbacks, and activitySession cleanup. Treat this as promising validation evidence for the tested run, not final consumer validation.
• May 12 branch validation reported that explicit UIKit host mode still preserves dismissal-completion behavior and delegate callbacks/session cleanup remain after confirmed dismissal. Broader compile validation still depends on resolving the XFlowViewMaker / XFlowSDK dependency mismatch.
• May 13 AccountLink runtime validation looks good in both SwiftUI-default and forced UIKit-host modes. Both paths preserved the intended dismissal sequencing, with delegate callbacks and session cleanup after confirmed dismissal.
• May 14 SampleApp work refined dismissal validation coverage: SampleApp can exercise both UIKit-host and SwiftUI-host paths, while XFlowSDK entrypoints prepare the matching dismissal mechanism for the active integration path. This supports validating that UIKit dismiss completion and SwiftUI XFlowDismissalHost lifecycle sequencing both converge on the canonical delegate/session teardown path.
• Jeff recommended broad Fid4 validation before PR review: test as many current XFlow entry and dismissal points as possible in both host modes, especially AO flows, and use temporary logs to confirm the active entry path, host mode, dismissal mechanism, delegate callbacks, and session cleanup.
• Quy confirmed on May 13 that this story and PDIAP-12284 can both remain In Progress together, so no immediate Jira restructuring is required.
• Sequenced after PDIAP-15838 source work, but merge/release is delayed until after REST-transition consumer validation
• Sized at 8 points

---

Context

• This story came out of the AccountLink root cause investigation.
• It is tied to the dismissal sequencing problem found after UIKit removal in SwiftUI-only paths.
• The root cause document was updated and the revised wording was approved.

---

Approved Scope

• Modernize dismissal delegate lifecycle sequencing for pure SwiftUI flows.
• Cover the missing lifecycle contract where delegate callbacks can happen before the view is fully removed.
• Validate the change across affected SwiftUI flows rather than only in one narrow reproduction.
• Preserve a single canonical delegate/session-clear path if possible, and do not treat SwiftUI onDisappear as proof of dismissal completion unless simulator logs validate that lifecycle contract.

---

Sequencing And Dependencies

• This story should come after PDIAP-15838.
• PDIAP-12284 is the original UIKit-wrapping removal story and was reopened after rollback.
• Current working relationship: implement and validate PDIAP-15836 together with PDIAP-12284 in the same branch unless direction changes. David identified a possible isolated/minimal fix path, but Jeff prefers combined branch work because consumer testing is required either way.
• It is aligned with epic 26Q2 - Updating XFlowSDK to Decouple and Fix ApexKit Dependencies (Split Part 2).
• If possible, it should use the same consumer-impact feature flag strategy as the broader UIKit-removal rollout.
• Current combined-branch planning should keep dismissal sequencing host-agnostic: the SwiftUI host path must prove dismissal completion before delegate callbacks, while the temporary UIHostingController fallback should preserve legacy dismiss-completion behavior.
• Current validation logs did not show the key failure patterns for the tested run, such as delegate firing before host-disappearance confirmation or repeated host-disappearance confirmation for the same dismissal id.
• SampleApp should be used as a guardrail for stale host-mode state by running UIKit-host and SwiftUI-host flows back-to-back without restarting the app.
• Fid4 validation should also use the previous Confluence entry-point list as a starting point, but current source inspection/logging should decide which entries are still valid, renamed, removed, or replaced.
• Expect a long-lived branch: after implementation, maintain the branch until consumer-testing approval. Jeff expects the GraphQL-removal branch to merge first after the REST validation period, then that branch should be merged into the PDIAP-15836 / PDIAP-12284 branch. Current estimate is roughly 90-100 days from 2026-05-05 unless Fidelity shortens the review windows.

---

Communication Notes

• Keep the scope framed as lifecycle sequencing in a pure SwiftUI environment, not only as a symptom like multiple modal presentation.
• If mentioned externally, keep it separate from PDIAP-15838.
• The Confluence/root-cause document was updated to reflect that FTTransfer changes are not the primary need anymore.
• The primary recommendation is to adjust the dismissal handling/sequencing correctly in the hosting path.
• FTTransfer improvements can still be mentioned as a secondary improvement, but not as a required change to reproduce the same visual behavior.
• When explaining why code still remains, clarify that the spike identified and documented the root cause, but implementation changes have not yet been published.