perf(crowdnode): persist the fruitless-restore memo per wallet

[QuantumExplorer]

Issue being fixed or feature implemented

CrowdNode.restoreState() runs a full post-2022 history scan when looking for a CrowdNode account. The scan is memoized in-process when it finds nothing (fruitlessRestoreTxCount), but that memo died with the process — so every cold launch of a wallet that never used CrowdNode still paid one ~7s main-thread scan at sync-done (measured ~1.1 ms/row materialization on a 7k-tx store).

What was done?

Persisted the memo per wallet in CrowdNodeDefaults (already scoped by active walletId), so a launch skips the scan when a previous launch already scanned the exact same history fruitlessly:

• New per-wallet key crowdNodeFruitlessRestoreTxCountKey, cached like the other CrowdNodeDefaults properties. restoreState() reads/writes it in place of the old in-memory var (the prefs _-cache plays the old in-memory role).
• Keyed to the persisted-row count read before the scans — the same value the pass actually saw, never re-read after (a save landing mid-restore must not mask unscanned rows).
• Written only by the account-not-found branch. A nil row count (SDK container not up yet) writes nothing — the scans were vacuous, and a prior launch's valid memo must not be clobbered.
• Cleared everywhere a rescan must be guaranteed: resetUserDefaults() / reset() (network change, alien-address teardown), resetForWipe() and clearPerWalletKeys(forWalletIdHex:) via membership in kPerWalletKeys (full wipe, single-wallet deletion), and handleActiveWalletChanged() — the row count is store-global while the scan is wallet-scoped, so a memo can't be trusted across a wallet switch.
• A restored seed whose history syncs in later rescans naturally: new rows → count differs → memo miss.
• Key resolution deliberately bypasses resolvedKey's seed-from-legacy step: this key is post-multi-wallet (no legacy install base), and a bare-key value can only come from a no-active-wallet pass whose wallet-scoped scan was vacuous — seeding it would let a wallet skip a scan it never ran.

Orthogonal to the CrowdNode release posture: the state machine stays app/SDK-owned, no DashSync surface is touched, and the change is inert while CrowdNode is hidden.

How Has This Been Tested?

• Clean dashpay scheme build (arm64 iOS simulator).
• QA-iPhone16 simulator smoke against the synthetic 7k-tx mainnet store, three launches:
  • Launch A (no memo): one restore pass, full scan (5944 post-2022 rows, ~7s fetch), "account not found", memo persisted as exactly the store's row count (7007) under the active wallet's key.
  • Launch B (memo == count): reached the same checkCrowdNodeState trigger, zero "restoring CrowdNode state" / "CrowdNode scan:" log lines — cold-launch scan skipped.
  • Launch C (memo forced stale, 6900 ≠ 7007): full rescan ran — exercises the restored-seed path.
• Simulator restored to testnet + pre-synthetic store afterwards.

Unit-test target is currently broken repo-wide (pre-existing); verification standard per CLAUDE.md is the clean dashpay build + testnet/sim smoke above.

Breaking Changes

None.

Checklist:

• I have performed a self-review of my own code
• I have commented my code, particularly in hard-to-understand areas
• I have added or updated relevant unit/integration/functional/e2e tests
• I have made corresponding changes to the documentation

For repository code-owners and collaborators only

• I have assigned this pull request to a milestone

🤖 Generated with Claude Code

Summary by CodeRabbit

• Improvements
  • Restore progress is now saved separately for each wallet.
  • Reopening the app or switching wallets preserves the relevant restore state.
  • Restores can avoid unnecessary rescanning when transaction data has not changed.
  • Restore state is cleared appropriately when wallets are reset or removed, ensuring future restores use fresh data.
• Bug Fixes
  • Prevented valid restore information from being overwritten when transaction data is temporarily unavailable.

[coderabbitai]

Caution

Review failed

The pull request is closed.

ℹ️ Recent review info

⚙️ Run configuration

Configuration used: defaults

Review profile: CHILL

Plan: Pro Plus

Run ID: 2bd0c9e2-ab1d-45d5-a934-f921e1edf642

📥 Commits

Reviewing files that changed from the base of the PR and between 19c48d5 and 6ee68b0.

📒 Files selected for processing (2)

• DashWallet/Sources/Models/CrowdNode/CrowdNode+UserDefaults.swift
• DashWallet/Sources/Models/CrowdNode/CrowdNode.swift

---

📝 Walkthrough

Walkthrough

CrowdNode now persists the fruitless restore transaction count per wallet. Restore skips scans when the stored count matches the current count. Reset and active-wallet changes clear the stored memo.

Changes

Per-wallet fruitless restore persistence

Layer / File(s)	Summary
Memo storage and lifecycle DashWallet/Sources/Models/CrowdNode/CrowdNode+UserDefaults.swift	Adds wallet-scoped storage for the fruitless restore count. Cache invalidation, wallet deletion, full wipes, and active-wallet resets clear the value.
Restore flow integration DashWallet/Sources/Models/CrowdNode/CrowdNode.swift	Uses the persisted count to skip unchanged restores. Records the pre-scan count only when available and clears the memo during resets and active-wallet changes.

Estimated code review effort: 3 (Moderate) | ~20 minutes

Possibly related PRs

• dashpay/dashwallet-ios#931: Introduces the CrowdNode fruitless-restore memoization extended by this change.
• dashpay/dashwallet-ios#973: Also modifies CrowdNode.restoreState() and restore scan memoization.

Suggested reviewers: llbartekll, jeanpierreroma

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch claude/youthful-heyrovsky-85daf2

---

_{Comment @coderabbitai help to get the list of available commands.}