Deleting a server record does not automatically remove an old offline copy or stop a queued edit from recreating it. Define how devices learn about removal and how stale operations are handled when they return.
Separate hiding a record, deleting active data, and retaining a removal marker for synchronisation. Those states can have different lifecycles and policy requirements.
Preserve enough information to reject stale resurrection
A device needs a reliable way to learn that a record was removed rather than merely absent from one response. The design may use deletion markers, revision history, a full refresh boundary, or another explicit protocol.
Choose how long removal knowledge remains available relative to supported offline periods. If the protocol forgets a deletion before an old device returns, it needs another safe reconciliation path instead of treating every stale record as new.
Reconnect a fictional deleted request
Suppose a tablet holds request version 8 and edits it offline. A phone deletes the request while online. When the tablet reconnects, the product must apply a defined rule: reject the edit and preserve a reviewable draft, restore through an authorised action, or another intentional outcome.
Test both arrival orders, a repeated deletion, a deleted parent with pending attachments, and a device that missed several sync cycles. Inspect local visibility, queue state, server records, and storage cleanup.
Do not silently resurrect the request because an update endpoint doubles as a create endpoint. Conversely, do not discard irreplaceable unsent notes without the policy and interface explaining what happened.
Include files, search, and account changes
Removal should reach relevant cached views, exports, attachment access, and background operations under the defined lifecycle. A list that hides the request while its file remains accessible may not fulfil the intended promise.
Current permission still matters. A former member returning with an offline copy should not receive fresh private data merely to complete reconciliation. Decide what local cleanup and user communication are appropriate under the organisation's policy.
The tradeoff is sync history versus retained metadata and complexity. Long-lived removal markers help stale devices but require storage, access, and retention decisions. A bounded offline-support window can simplify the protocol if it is explicit and safe for users outside that window.
Flutter's offline-first guide supplies architectural context. The Social App portfolio account documents related work. This request lifecycle is fictional and makes no compliance or implementation claim.
Questions about offline deletion
Is an absent server row enough information?
Not always. The client may need to distinguish deletion from filtering, permission loss, incomplete results, or a temporary failure.
Should a stale edit recreate deleted data?
Only through an explicit permitted policy. Accidental resurrection is not a conflict-resolution strategy.
What should be tested first?
Delete online, edit offline elsewhere, then reconnect after a long controlled gap. Bring the resulting records and retained draft to a sync-lifecycle review.