fix: classify notification reply send failures

[agent-p1p]

Summary

• Return a typed notification-reply send outcome instead of collapsing every failure to false.
• Classify only proven pre-send relay connection failures as transient; invalid, terminal, and ambiguous post-send failures fail fast.
• Preserve the existing three-attempt WorkManager bound for transient failures and cancellation propagation.

Test plan

• ./gradlew :app:testDebugUnitTest --tests dev.ipf.whitenoise.android.state.TransientRelaySendErrorTest --tests dev.ipf.whitenoise.android.notifications.NotificationTargetTest
• ./gradlew :app:testDevDebugUnitTest :app:lintDevDebug --no-daemon --stacktrace
• ./gradlew :app:verifyRoborazziDevDebug --no-daemon --stacktrace
• ./gradlew :app:ktlintCheck --no-daemon --stacktrace

Fixes #1276

---

Summary by CodeRabbit

• Bug Fixes

  • Improved notification reply delivery handling by distinguishing successful, temporary, and permanent failures.
  • Temporary delivery issues are retried automatically, while permanent failures stop without unnecessary retries.
  • Successfully sent replies now consistently mark messages as read, dismiss related notifications, and record the action as completed.
  • Invalid reply details are handled safely without triggering retries.

• Tests

  • Added coverage for retry limits and failure classification.

[coderabbitai]

Note

Currently processing new changes in this PR. This may take a few minutes, please wait...

⚙️ Run configuration

Configuration used: Organization UI

Review profile: CHILL

Plan: Pro

Run ID: 4c4388d6-fd4b-41c8-97e9-2c390f46673d

📥 Commits

Reviewing files that changed from the base of the PR and between 58fb054 and d1d8ce2.

📒 Files selected for processing (4)

• app/src/main/java/dev/ipf/whitenoise/android/notifications/NotificationReplyWorker.kt
• app/src/main/java/dev/ipf/whitenoise/android/state/AppState.kt
• app/src/test/java/dev/ipf/whitenoise/android/notifications/NotificationTargetTest.kt
• app/src/test/java/dev/ipf/whitenoise/android/state/TransientRelaySendErrorTest.kt

✨ 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 pip/whitenoise-android-1276

---

Thanks for using CodeRabbit! It's free for OSS, and your support helps us grow. If you like it, consider giving us a shout-out.

❤️ Share

• X
• Mastodon
• Reddit
• LinkedIn

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

[stage-review]

Ready to review this PR? Stage has broken it down into 4 individual chapters for you:

	Title
1	Define notification reply send outcomes
2	Update AppState to return typed outcomes
3	Integrate typed outcomes into NotificationReplyWorker
4	Verify worker retry and result logic

_{Chapters generated by Stage for commit d1d8ce2 on Jul 14, 2026 6:11pm UTC.}

[github-actions]

Preview APK for PR #1378

Home-screen label: 1378 White Noise with the blueprint launcher icon — installs side-by-side with the production and staging apps.

Install directly — one tap, no GitHub sign-in (public, content-addressed .apk). ARM64 only.

Every push to this PR rebuilds and updates this comment.

[agent-p1p]

Pip adversarial review for #1276 at 85d7b0d6.

BLOCKING:

• none

SUGGESTION:

• app/src/main/java/dev/ipf/whitenoise/android/state/AppState.kt:288-304 — PERMANENT_FAILURE is broader than its name: the classifier intentionally puts both truly permanent errors and retry-unsafe/ambiguous failures into this bucket. For example, MarmotKitException.StorageBusy is typed as transient by the FFI but can surface after publication during local projection/persistence, where an automatic resend could duplicate the message. The current fail-fast behavior is correct; renaming this outcome to NON_RETRYABLE_FAILURE or FAIL_FAST would encode the actual invariant and discourage unsafe broadening later.
• app/src/main/java/dev/ipf/whitenoise/android/notifications/NotificationReplyWorker.kt:64-79 — focused tests cover the pure retry predicate, but not the doWork() outcome-to-Result wiring. A worker-level test for SENT, permanent/non-retryable, and transient outcomes would guard the cross-layer mapping this issue adds. Non-blocking because the mapping is small and the existing test style uses pure seams.

NITPICK:

• none

SUMMARY: The typed outcome fixes the Boolean collapse. Cancellation still propagates, permanent and retry-unsafe failures fail immediately, and only the existing narrow connect-phase classifier enters the bounded three-attempt WorkManager retry path. The raw notification send does not pass through the foreground controller retry loop, so there is no double retry. Focused verification passed 43/43 tests; all GitHub checks are green. No sensitive paths were touched.

VERDICT: APPROVE
SENSITIVE_PATHS: none

[agent-p1p]

Addressed both adversarial-review suggestions in d1d8ce2: renamed the fail-fast bucket to NON_RETRYABLE_FAILURE and added direct worker result-mapping coverage for sent, non-retryable, retryable, and exhausted-retry outcomes. Local full gates and fresh GitHub CI pass.

[agent-p1p]

@coderabbitai review

[coderabbitai]

✅ Action performed

Review finished.

Note: CodeRabbit is an incremental review system and does not re-review already reviewed commits. This command is applicable only when automatic reviews are paused.