fix(spv): add distinct error state for SPV sync failures#650
Merged
Conversation
The SpvManager's progress watcher and sync event handler did not detect fatal sync errors, leaving the UI stuck in "Syncing" status indefinitely. Two detection paths are now in place: 1. spawn_sync_event_handler handles SyncEvent::ManagerError to transition SpvStatus to Error and store the error message. This is the primary path since dash-spv always emits this event on manager failure. 2. spawn_progress_watcher checks SyncState::Error from the progress channel as defense-in-depth (currently blocked by upstream bug dashpay/rust-dashcore#469 where try_emit_progress is not called on error paths). Once SpvStatus::Error is set, the existing ConnectionStatus logic maps it to OverallConnectionState::Disconnected, showing a red indicator icon with "SPV: Error" in the tooltip. Related upstream issues: - dashpay/rust-dashcore#469 (missing progress emit on error) - dashpay/rust-dashcore#470 (QRInfo chain lock retry robustness) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Contributor
📝 WalkthroughWalkthroughSurfaces SPV sync errors into application state and UI: introduces OverallConnectionState::Error and spv_last_error storage, records last_error in the SPV manager (progress watcher and ManagerError), updates connection tooltips, top-panel indicator (magenta + “!”), theme color, and adds a manual test scenarios doc. Changes
Sequence Diagram(s)sequenceDiagram
rect rgba(200,200,255,0.5)
participant SPV as SPV Library
end
rect rgba(200,255,200,0.5)
participant Manager as SPV Manager
end
rect rgba(255,200,200,0.5)
participant Status as ConnectionStatus
end
rect rgba(255,230,200,0.5)
participant UI as TopPanel/UI
end
SPV->>Manager: progress update (SyncState)
alt SyncState::Error
Manager->>Manager: set last_error (Arc) if not present
Manager->>Status: update SpvStatus::Error with detail
Manager->>Status: emit diagnostic log
Status->>UI: refresh -> OverallConnectionState::Error (include detail)
UI->>UI: render magenta indicator + "!"
else SyncState::Running/Complete
Manager->>Status: update SpvStatus::Running/Synced
Status->>UI: refresh -> Synced/Syncing
end
Estimated code review effort🎯 3 (Moderate) | ⏱️ ~20 minutes Possibly related issues
Poem
🚥 Pre-merge checks | ✅ 3✅ Passed checks (3 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches
🧪 Generate unit tests (beta)
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. Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/spv/manager.rs`:
- Around line 1082-1087: The progress-watcher branch currently unconditionally
sets last_error to "Sync failed (reported by SPV library)" when is_error is
true, which can overwrite a more detailed error previously set by the sync event
handler; change the logic in the is_error branch (where SpvStatus::Error is
assigned) to open last_error.write(), inspect whether it is None, and only
assign the generic message if err_guard.is_none(), leaving an existing detailed
message intact (keep the existing status_guard = SpvStatus::Error assignment).
ℹ️ Review info
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (2)
docs/ai-design/2026-02-24-spv-sync-error-status/manual-test-scenarios.mdsrc/spv/manager.rs
Only set the generic "Sync failed" message in last_error when no detailed message has already been stored by the sync event handler. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Contributor
There was a problem hiding this comment.
♻️ Duplicate comments (1)
src/spv/manager.rs (1)
1082-1089: Past concern resolved —is_none()guard correctly preserves detailed error message.The
err_guard.is_none()check ensures the generic fallback message does not overwrite a detailed message already stored by the sync event handler, which is the fix requested in the prior review.🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/spv/manager.rs` around lines 1082 - 1089, The code correctly avoids overwriting a detailed error by checking last_error.write() and err_guard.is_none() before setting a generic message; no code change needed—ensure the conditional around SpvStatus::Error uses the same pattern (is_error branch setting *status_guard = SpvStatus::Error and only writing to last_error when err_guard.is_none()) to preserve previously stored detailed errors from the sync event handler.
🧹 Nitpick comments (1)
src/spv/manager.rs (1)
1082-1089: Consider adding inline unit tests for the new error-transition paths.Neither the
is_errorbranch inspawn_progress_watchernor theSyncEvent::ManagerErrorhandler inspawn_sync_event_handlerhas test coverage. The status/last_error mutation logic can be tested independently of the async spawn by extracting it into a small helper or usingtokio::test. As per coding guidelines, unit tests should be written inline using#[test].Also applies to: 1155-1163
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/spv/manager.rs` around lines 1082 - 1089, Add inline unit tests covering the error-transition logic by extracting the status/last_error mutation into a small helper (e.g., set_spv_error_status or update_status_on_error) and then write #[test] (or #[tokio::test] if async) cases that exercise the is_error branch used by spawn_progress_watcher and the SyncEvent::ManagerError handling used by spawn_sync_event_handler; tests should call the helper (or invoke the handler functions directly) to assert SpvStatus becomes SpvStatus::Error and last_error is set to Some(...) when an error condition is simulated, and also include a case where last_error is already Some to ensure it is not overwritten.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Duplicate comments:
In `@src/spv/manager.rs`:
- Around line 1082-1089: The code correctly avoids overwriting a detailed error
by checking last_error.write() and err_guard.is_none() before setting a generic
message; no code change needed—ensure the conditional around SpvStatus::Error
uses the same pattern (is_error branch setting *status_guard = SpvStatus::Error
and only writing to last_error when err_guard.is_none()) to preserve previously
stored detailed errors from the sync event handler.
---
Nitpick comments:
In `@src/spv/manager.rs`:
- Around line 1082-1089: Add inline unit tests covering the error-transition
logic by extracting the status/last_error mutation into a small helper (e.g.,
set_spv_error_status or update_status_on_error) and then write #[test] (or
#[tokio::test] if async) cases that exercise the is_error branch used by
spawn_progress_watcher and the SyncEvent::ManagerError handling used by
spawn_sync_event_handler; tests should call the helper (or invoke the handler
functions directly) to assert SpvStatus becomes SpvStatus::Error and last_error
is set to Some(...) when an error condition is simulated, and also include a
case where last_error is already Some to ensure it is not overwritten.
Distinguish "connected but sync failed" from "never connected" by adding an `OverallConnectionState::Error` variant. SPV sync errors now show a magenta pulsating circle with "!" glyph instead of the same red static circle used for disconnected state. Tooltip shows the specific SPV error message for easier debugging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
lklimek
changed the title
- Move last_error write outside status write guard in spawn_progress_watcher to maintain consistent lock ordering (status → release → last_error), eliminating latent deadlock risk. - Update manual test scenarios to match actual implementation: magenta pulsating "!" indicator (not red/static/Disconnected). - Add Scenario 4 to explicitly verify Error vs Disconnected distinction. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Contributor
Author
Audit SummaryReviewed by: Claude Code with a 2-agent team:
Overall Risk: LOW — well-scoped status propagation change with no new attack surface. Findings
Pre-existing / Out-of-scope
Positive Observations
RedundancyBoth agents flagged the nested lock issue (code-reviewer as "overwrite asymmetry", security as "nested lock acquisition"). Security's framing was more precise — the lock ordering is the root concern. Redundancy ratio: 1 finding out of 8 unique = 12.5%. Verdict: Approve. The two MEDIUM findings have been fixed. Remaining LOW/INFO items are acceptable trade-offs for a desktop application. 🤖 Co-authored by Claudius the Magnificent AI Agent |
…r-status # Conflicts: # src/context/connection_status.rs
Contributor
There was a problem hiding this comment.
🧹 Nitpick comments (1)
src/ui/components/top_panel.rs (1)
117-125: Note: Error state pulsation runs indefinitely.The pulsation continues for the terminal Error state since it doesn't match
Disconnected. This causes continuous repaints viarepaint_animation()at line 167. Per PR audit this is noted as a LOW-priority concern (minor CPU/GPU usage).If desired, you could stop the pulsation for Error state by adding it to the non-pulsating check:
OverallConnectionState::Disconnected | OverallConnectionState::Error => 1.0,But this is optional since the visual feedback of a pulsating error indicator may be intentional UX.
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/ui/components/top_panel.rs` around lines 117 - 125, The Error state is currently treated as pulsating because pulse_scale's match covers OverallConnectionState::Error with a sinusoidal value, causing continuous repaints via repaint_animation(); to stop pulsation for terminal Error state (like Disconnected) change the match arm for pulse_scale so Error maps to the constant 1.0 instead of a sine expression (update the match in the code that computes pulse_scale for OverallConnectionState), ensuring repaint_animation() no longer receives continuous animation-triggering updates when state is Error.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Nitpick comments:
In `@src/ui/components/top_panel.rs`:
- Around line 117-125: The Error state is currently treated as pulsating because
pulse_scale's match covers OverallConnectionState::Error with a sinusoidal
value, causing continuous repaints via repaint_animation(); to stop pulsation
for terminal Error state (like Disconnected) change the match arm for
pulse_scale so Error maps to the constant 1.0 instead of a sine expression
(update the match in the code that computes pulse_scale for
OverallConnectionState), ensuring repaint_animation() no longer receives
continuous animation-triggering updates when state is Error.
Contributor
Author
🔍 Audit Summary — PR #650Reviewed by: Claude Code with a 2-agent team:
Overall AssessmentClean, well-scoped PR that fills a real UX gap. No critical or high-severity issues. The concurrency model is sound, match arms are exhaustive, and the error state lifecycle (create → display → cleanup) is handled correctly across all three touchpoints (manager, connection status, UI). Findings
Pre-existing / Out-of-scope
Positive Observations
🤖 Reviewed by Claudius the Magnificent AI Agent |
…llocations - Add explicit drop(guard) in spawn_sync_event_handler for lock ordering - Use first-error-wins policy consistently, log subsequent errors - Add failed_manager_name() helper for detailed progress-channel errors - Add TODO for error string truncation (CWE-400) - Use Cow<str> in SPV tooltip to avoid unnecessary allocations - Document Mutex choice for spv_last_error - Update animation repaint comment to list all pulsating states Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Contributor
Author
🔍 Re-Audit Summary — PR #650 (Round 2)Reviewed by: Claude Code with a 2-agent team:
Overall AssessmentAll 7 findings from the first review are properly addressed. Security engineer approves for merge. One new MEDIUM found by the Rust reviewer (tooltip inconsistency), plus minor items. New Findings
Finding #1 DetailWhen Fix: Add an explicit branch for Verification of Previous Fixes
Positive Observations
Redundancy ratio: 0/5 new findings overlapped between agents (disjoint focus areas this round). 🤖 Reviewed by Claudius the Magnificent AI Agent |
- Add explicit SpvStatus::Error branch in spv_label to show "SPV: Error"
instead of falling through to spv_phase_summary() which shows "syncing..."
- Use Display ({}) instead of Debug ({:?}) for ManagerIdentifier in tracing
- Document masternodes-first order in failed_manager_name()
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
PastaPastaPasta
approved these changes
Feb 27, 2026
…r-status # Conflicts: # src/context/connection_status.rs # src/ui/components/top_panel.rs
Contributor
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
src/context/connection_status.rs (1)
113-115: Hardenspv_last_errorlocking and snapshot assignment semantics.At lines 113, 321–326, and 452–455, lock failures silently skip updates via
if let Ok()patterns, which can leave stale error state. Additionally, at line 452, movingsnapshot.last_errorintoself.spv_last_erroris fragile when the field is non-Copy(it isOption<String>); clone explicitly for safety.♻️ Suggested hardening
- if let Ok(mut err) = self.spv_last_error.lock() { - *err = None; - } + let mut err = self.spv_last_error.lock().unwrap_or_else(|e| e.into_inner()); + *err = None; ... - let detail = self - .spv_last_error - .lock() - .ok() - .and_then(|g| g.clone()) - .unwrap_or_else(|| "unknown error".to_string()); + let detail = self + .spv_last_error + .lock() + .unwrap_or_else(|e| e.into_inner()) + .clone() + .unwrap_or_else(|| "unknown error".to_string()); ... - if let Ok(mut err) = self.spv_last_error.lock() { - *err = snapshot.last_error; - } + let mut err = self.spv_last_error.lock().unwrap_or_else(|e| e.into_inner()); + *err = snapshot.last_error.clone();🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed. In `@src/context/connection_status.rs` around lines 113 - 115, The current code silently ignores PoisonError/lock failures when updating self.spv_last_error (seen around spv_last_error lock sites) and moves snapshot.last_error into the field, which is fragile for Option<String>; update each if let Ok(self.spv_last_error.lock()) pattern to explicitly handle Err by logging or returning an error (do not silently skip the update), and when assigning from snapshot.last_error use an explicit clone (e.g., snapshot.last_error.clone()) to avoid moving non-Copy data; locate and change the three sites that lock spv_last_error (the blocks around the existing if let Ok(...) at lines ~113, ~321–326, and ~452–455) to use match on the lock result and perform a clear assignment of Some/None with cloned string data.
🤖 Prompt for all review comments with AI agents
Verify each finding against the current code and only fix it if needed.
Inline comments:
In `@src/ui/components/top_panel.rs`:
- Around line 168-170: The comment above the repaint call is out of sync with
the runtime check: the code calls app_context.repaint_animation(ui.ctx()) for
any OverallConnectionState that is not OverallConnectionState::Disconnected
(including Connecting), so update the comment to reflect that repaint is
requested for all non-Disconnected states (Connecting, Synced, Syncing, Error)
or rephrase to "Request repaint for all non-Disconnected connection states" next
to the repaint call referencing OverallConnectionState and
app_context.repaint_animation/ui.ctx() to keep intent clear.
---
Nitpick comments:
In `@src/context/connection_status.rs`:
- Around line 113-115: The current code silently ignores PoisonError/lock
failures when updating self.spv_last_error (seen around spv_last_error lock
sites) and moves snapshot.last_error into the field, which is fragile for
Option<String>; update each if let Ok(self.spv_last_error.lock()) pattern to
explicitly handle Err by logging or returning an error (do not silently skip the
update), and when assigning from snapshot.last_error use an explicit clone
(e.g., snapshot.last_error.clone()) to avoid moving non-Copy data; locate and
change the three sites that lock spv_last_error (the blocks around the existing
if let Ok(...) at lines ~113, ~321–326, and ~452–455) to use match on the lock
result and perform a clear assignment of Some/None with cloned string data.
ℹ️ Review info
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Pro
📒 Files selected for processing (3)
src/context/connection_status.rssrc/spv/manager.rssrc/ui/components/top_panel.rs
🚧 Files skipped from review as they are similar to previous changes (1)
- src/spv/manager.rs
Comment on lines
+168
to
170
| // Request repaint for animation (Synced, Syncing, and Error states pulse) | ||
| if overall != OverallConnectionState::Disconnected { | ||
| app_context.repaint_animation(ui.ctx()); |
Contributor
There was a problem hiding this comment.
Update repaint comment to match runtime condition.
At Line [169], repaint is requested for every non-Disconnected state, including Connecting, but the comment mentions only Synced/Syncing/Error.
📝 Suggested comment fix
- // Request repaint for animation (Synced, Syncing, and Error states pulse)
+ // Request repaint for animation in all non-disconnected states📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| // Request repaint for animation (Synced, Syncing, and Error states pulse) | |
| if overall != OverallConnectionState::Disconnected { | |
| app_context.repaint_animation(ui.ctx()); | |
| // Request repaint for animation in all non-disconnected states | |
| if overall != OverallConnectionState::Disconnected { | |
| app_context.repaint_animation(ui.ctx()); |
🤖 Prompt for AI Agents
Verify each finding against the current code and only fix it if needed.
In `@src/ui/components/top_panel.rs` around lines 168 - 170, The comment above the
repaint call is out of sync with the runtime check: the code calls
app_context.repaint_animation(ui.ctx()) for any OverallConnectionState that is
not OverallConnectionState::Disconnected (including Connecting), so update the
comment to reflect that repaint is requested for all non-Disconnected states
(Connecting, Synced, Syncing, Error) or rephrase to "Request repaint for all
non-Disconnected connection states" next to the repaint call referencing
OverallConnectionState and app_context.repaint_animation/ui.ctx() to keep intent
clear.
lklimek
added a commit
that referenced
this pull request
Mar 2, 2026
Cover the new Error variant added by #650 (SPV sync error state) in the connection banner match arm. Shows an error banner directing users to the connection status tooltip for details. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
lklimek
added a commit
that referenced
this pull request
Mar 3, 2026
Semantic merge conflict from v1.0-dev: PR #650 added an Error variant to OverallConnectionState, which our connection banner match didn't cover. Show an error banner when SPV sync enters the error state. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
lklimek
added a commit
that referenced
this pull request
Mar 3, 2026
* docs: add unified message display design documents Add UX specification, technical architecture, and HTML mockup for the MessageBanner component that will replace the ~50 ad-hoc error/message display implementations across screens with a single reusable component. Key design decisions: - Per-screen MessageBanner with show()/set_message() API - All colors via DashColors (zero hardcoded Color32 values) - 4 severity levels: Error, Warning, Success, Info - Auto-dismiss for Success/Info (5s), persistent for Error/Warning - Follows Component Design Pattern conventions (private fields, builder, show) - No changes to BackendTask/TaskResult/AppState architecture Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add MessageType::Warning * chore: initial implementation * chore: docs * chore: self review * refactor(context): replace RwLock<Sdk> with ArcSwap<Sdk> (#600) * refactor(context): replace RwLock<Sdk> with ArcSwap<Sdk> Sdk is internally thread-safe (Arc, ArcSwapOption, atomics) and all methods take &self. The RwLock was adding unnecessary contention across backend tasks. Using ArcSwap instead of plain Sdk because reinit_core_client_and_sdk() needs to atomically swap the entire Sdk instance when config changes. ArcSwap provides lock-free reads with atomic swap for the rare write. Suggested-by: lklimek * fix: address CodeRabbit review findings for ArcSwap migration - Fix import ordering: move arc_swap::ArcSwap before crossbeam_channel - Remove redundant SDK loads in load_identity_from_wallet, register_dpns_name, and load_identity — use the sdk parameter already passed to these functions - Fix stale TODO referencing removed sdk.read().unwrap() API - Rename sdk_guard → sdk in transfer, withdraw_from_identity, and refresh_loaded_identities_dpns_names (no longer lock guards) - Pass &sdk to run_platform_info_task from dispatch site instead of reloading internally - Fix leftover sdk.write() call in context_provider.rs (RwLock remnant) - Add missing Color32 import in wallets dialogs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: address remaining CodeRabbit review feedback on ArcSwap migration - Move SDK load outside for loop in refresh_loaded_identities_dpns_names.rs so it's loaded once for the batch instead of on each iteration - Update stale TODO comment in default_platform_version() to reflect that this is a free function with no sdk access Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: consolidate double read-lock on spv_context_provider Clone the SPV provider in a single lock acquisition, then bind app context on the clone instead of locking twice. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: PastaClaw <thepastaclaw@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> * refactor: remove unused Insight API and show_in_ui config fields (#597) * refactor: remove unused Insight API references The `insight_api_url` field in `NetworkConfig` and its associated `insight_api_uri()` method were never used in production code (both marked `#[allow(dead_code)]`). Remove the field, method, config entries, env example lines, and related tests. https://claude.ai/code/session_01HWPmCJHT8KTZGP9bFiksjn * refactor: remove unused `show_in_ui` field from NetworkConfig The `show_in_ui` field was defined on `NetworkConfig` and serialized in `save()`, but never read by any production code to control network visibility. Remove the field, its serialization, env example lines, and test references. https://claude.ai/code/session_01HWPmCJHT8KTZGP9bFiksjn * fix: add missing `Color32` import in wallet dialogs https://claude.ai/code/session_01HWPmCJHT8KTZGP9bFiksjn --------- Co-authored-by: Claude <noreply@anthropic.com> * build: add Flatpak packaging and CI workflow (#589) * build: remove snap version * build: add Flatpak packaging and CI workflow Add Flatpak build manifest, desktop entry, AppStream metadata, and GitHub Actions workflow for building and distributing Flatpak bundles. Uses freedesktop 25.08 runtime with rust-stable and llvm21 extensions. No application source code changes required - works in SPV mode by default. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: address review findings for Flatpak packaging - Pin GitHub Actions to commit SHAs for supply chain security - Upgrade softprops/action-gh-release from v1 to v2.2.2 - Remove redundant --socket=x11 (fallback-x11 suffices) - Remove duplicate tag trigger preventing double builds on release - Remove duplicate env vars inherited from top-level build-options - Add Flatpak build artifacts to .gitignore - Add bugtracker URL to AppStream metainfo - Remove deprecated <categories> from metainfo (use .desktop instead) - Add Terminal=false and Keywords to desktop entry - Add disk space check after SDK install in CI - Rename artifact to include architecture suffix Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: simplify CI workflows for Linux-only releases - Remove "Free disk space" step from flatpak and release workflows - Remove Windows and macOS builds from release workflow - Use "ubuntu" runner image instead of pinned versions - Clean up unused matrix.ext references Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: attach to existing releases instead of creating new ones - Replace release-creating job with attach-to-release (only on release event) - Add 14-day retention for build artifacts - On tag push or workflow_dispatch, only upload artifacts (no release) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * revert: restore release.yml to original v1.0-dev version The release workflow changes were out of scope for the Flatpak PR. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: address CodeRabbit review comments - Fix CRLF line endings in Flatpak manifest (convert to LF) - Set app_id on ViewportBuilder to match desktop StartupWMClass - Use --locked flag for reproducible cargo builds in Flatpak - Rename --repo=repo to --repo=flatpak-repo to match .gitignore - Add architecture note for protoc x86_64 binary Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs: add Flatpak install instructions to README Add a dedicated section for installing via Flatpak on Linux, clarify that prerequisites are only needed for building from source, and rename "Installation" to "Build from Source" for clarity. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: match StartupWMClass to Flatpak app_id Use reverse-DNS format org.dash.DashEvoTool to match the Wayland app_id set via ViewportBuilder::with_app_id(). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: use ** glob for branch trigger to match feat/flatpak Single * doesn't match path separators in GitHub Actions branch filters. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add aarch64 Flatpak build and caching to CI - Add matrix strategy for parallel x86_64 and aarch64 builds - Patch protoc URL/sha256 per architecture at build time - Cache .flatpak-builder directory keyed on arch + manifest + lockfile - Pin actions/cache to SHA Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: convert desktop and metainfo files to LF line endings Flatpak builder validates desktop files and rejects CRLF line endings. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: cancel in-progress Flatpak builds on new push Add concurrency group keyed on git ref so a new push cancels any running build for the same branch or release. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: address PR review findings for Flatpak packaging - Remove unnecessary --filesystem=xdg-config/dash-evo-tool:create (Flatpak already redirects XDG_CONFIG_HOME to sandbox) - Add categories and keywords to AppStream metainfo for discoverability - Update README with both x86_64/aarch64 install commands, uninstall instructions, and Flatpak data path note - Clarify aarch64 comment in manifest to reference CI sed patching Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: workflow timeout and perms * fix: move permissions to job level in Flatpak workflow Step-level permissions are not valid in GitHub Actions. Move contents: write to the job level where it is needed for release attachment. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: cache Cargo registry and target in Flatpak CI Bind-mount host-side cargo-cache and cargo-target directories into the Flatpak build sandbox so CARGO_HOME and target/ persist across builds. Uses split restore/save with cleanup of incremental and registry/src (similar to Swatinem/rust-cache). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: scope cargo cache bind-mount sed to build-args only The previous sed matched --share=network in both finish-args and build-args, corrupting finish-args. Use a sed range to only target the build-args section. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Apply suggestions from code review --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> * chore: fix build * chore: use new error handling everywhere - not self reviewed * chore: use message banner to show progress * fix: start elapsed counter at 1s instead of 0s Aligns elapsed display with the countdown timer which already adds 1 to avoid showing "0s" immediately. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: rabbit review * Update src/app.rs Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * chore: peer review * chore: fix build errors * fix(ui): make MessageBanner::set_global truly idempotent set_global() no longer resets timestamps, auto-dismiss timer, or logged flag when a banner with identical text already exists. This makes it safe to call every frame without log spam or timer restarts. Cherry-picked from origin/fix/spv-peer-timeout (08e3b3b). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address code review findings R01, R03, R06, R09, R10 - R01: Replace expect("No key selected") with graceful match + error banner in 11 token screens to prevent panics on missing signing key - R03: Remove dead backend_message field from AddExistingIdentityScreen - R06: Replace is_some() + unwrap() with idiomatic if-let-Some pattern in 10 token screens; use is_some_and() in structs.rs - R09: Add use imports for MessageBanner in 5 dashpay screens, replacing 22 fully-qualified crate::ui::components::MessageBanner:: calls - R10: Replace custom_dash_qt_error_message inline rendering with MessageBanner::set_global in network_chooser_screen Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address review findings SEC-08, SEC-09, RUST-015, SEC-05, SEC-07 - SEC-08: Restore safe if-let-Some pattern in WithdrawalScreen::refresh() to prevent double unwrap() panic on DB error or deleted identity - SEC-09: Restore original DB lookup in SendPaymentScreen::load_contact_info() replacing hardcoded "alice.dash" mock data - RUST-015: Revert unimplemented!() back to ui.label() in update_token_config MarketplaceTradeMode arm - SEC-05: Add success banners for contact request accept/reject in ContactRequests::display_task_result - SEC-07: Add MessageBanner::clear_all_global() and call it from AppState::change_network() to prevent stale banners leaking across network switches Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs: update coding conventions and message display guidance Add fallible constructor rule (Result<Self, ...> when they can fail), rename section to "General rules", and document MessageBanner idempotency (no guard needed for set_global). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): replace expect/panic with graceful error handling (SEC-10) Replace all expect() calls in token screen constructors and confirmation handlers with MessageBanner error display. Constructors handle errors internally and return Self with degraded state, keeping create_screen() clean. refresh() methods now show errors via MessageBanner instead of tracing-only logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): accept impl Display/Debug in MessageBanner API Change MessageBanner public methods to accept `impl Display` for message text and `impl Debug` for details, instead of `&str`. Remove needless `&format!(...)` borrows across 27 call sites. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): remove error Failed to get best chain lock for mainnet, testnet, devnet, and local Fixes #633 * feat(ui): add automatic connection status banners Display persistent MessageBanner notifications based on network connection state transitions. Mode-aware messages guide users toward the right recovery action (RPC vs SPV). - Disconnected (RPC): "Disconnected — check that Dash Core is running" - Disconnected (SPV): "Disconnected — check your internet connection" - Syncing (RPC): "Syncing with Dash Core…" - Syncing (SPV): "SPV sync in progress…" - Synced: banner cleared Uses Option<OverallConnectionState> for change detection, with None as initial/post-network-switch sentinel to force first evaluation. Closes #667 (partial — action links deferred to follow-up) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): pass TaskError directly to with_details() to avoid double-formatting The previous code used `format!("{err:?}")` which produced a String, then `with_details()` applied `{:#?}` again — wrapping the output in quotes and escaping inner characters. Passing `&err` directly lets Debug format once. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): correct copy-paste error messages in token screens Replace "Burning" error messages that were copy-pasted from burn screen into freeze, destroy, and resume token screens with contextually correct messages. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): restore lost success/error messages in 5 screens Replace display_message() calls with MessageBanner::set_global() in screens where display_message() is now a side-effect-only handler and no longer displays messages directly. Affected screens: create_asset_lock_screen, wallets_screen (MineBlocks), address_table (export error), profile_screen (validation), contact_details. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): replace unwrap/expect with graceful error handling Replace double unwrap in transfer_screen refresh() with unwrap_or_else + MessageBanner error display, matching the pattern from withdraw_screen. SEC-002 tokens_screen skipped: the .expect() calls are only on compile-time embedded image data (include_bytes!) which is safe. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): migrate masternode_list_diff_screen to global MessageBanner Replace ~15 local ui_state.message assignments and custom render_message_banner() with MessageBanner::set_global() via the display_message() trait method. Remove the message field from UiState and the unused Color32 import. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): improve banner eviction logging and atomics - Upgrade BANNER_KEY_COUNTER from Relaxed to SeqCst ordering for future-proofing against multi-threaded usage - Log evicted banners at warn level in set_global() and replace_global() - Add comment explaining why show_global() always writes back Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: remove resolved TODO.md All items tracked in the unified message display TODO have been addressed or moved to the review findings. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(ui): add INTENTIONAL markers and API documentation - Document why with_details() accepts Debug (not Display): structured error context is more useful in diagnostic details panes - Document replace_global() fallback-to-add behavior as intentional - Add INTENTIONAL(SEC-003) marker for developer mode error details - Add INTENTIONAL(SEC-004) marker for BannerHandle Send+Sync safety Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): extract shared token validation and fix ordering - Add validate_signing_key() helper in tokens/mod.rs to eliminate duplicated signing key validation across 12 token screens - Move signing key validation BEFORE WaitingForResult state transition so users see immediate errors instead of loading spinner then error - Replace is_err()/unwrap() anti-pattern with idiomatic let-else blocks in freeze, mint, transfer, destroy_frozen_funds, unfreeze screens Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): return Result from get_selected_wallet Replace &mut Option<String> error out-parameter with idiomatic Result<Option<Arc<RwLock<Wallet>>>, String>. Update 26+ callsites across identity, token, DashPay, and contract screens. Callsite patterns: unwrap_or_else with MessageBanner for user-visible errors, unwrap_or(None) where errors were previously silently ignored. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): resolve duplicate imports and clippy warnings Remove duplicate MessageBanner imports in create_asset_lock_screen and wallets_screen/mod. Fix needless_borrows_for_generic_args clippy lints in profile_screen, transfer_screen, and wallets_screen. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * style: reorder imports in masternode_list_diff_screen Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address review findings from MessageBanner migration audit Apply 13 triaged review fixes plus 1 bug fix across 22 files: - Remove dead error state fields (backend_message, error_message, Error variant) - Replace .expect() panics with graceful fallback + MessageBanner in token screens - Fix missing MessageBanner::show_global() on contracts documents screen - Migrate DocumentActionScreen inline errors to MessageBanner - Replace unwrap_or(None) with error-reporting fallback in DashPay screens - Fix replace_global idempotency and use relaxed atomic ordering in banner - Extract shared set_error_banner helper for 8 token screens - Restore correct Some(0) wallet index semantics - Document BannerHandle lifecycle in CLAUDE.md Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: validate token description length before sending to Platform (#530) * fix: validate token description length before sending to Platform Descriptions must be either empty or 3-100 characters long. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(ui): validate token description by char count, not byte length String::len() counts UTF-8 bytes, causing multi-byte characters (CJK, emoji) to be miscounted against the 3–100 limit. Switch to chars().count() and update all UI labels to surface the minimum. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com> Co-authored-by: Lukasz Klimek <842586+lklimek@users.noreply.github.com> * refactor(ui): consolidate banner extension traits into message_banner Move BannerHandleExt and ResultBannerExt from banner_ext.rs into message_banner.rs where they belong. Merge take_and_clear() into OptionBannerExt trait (impl for Option<BannerHandle>) alongside or_show_error() for Option. Remove the separate banner_ext module and Clearable helper trait for simplicity. Apply review findings: DRY patterns (take_and_clear, or_show_error, load_identities_with_banner), fix .expect() panics in constructors, restore known_identities in refresh(), narrow pub field visibility, add ScreenLike doc comments, and update CLAUDE.md conventions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * doc(tmp): review guide for pr 604 * fix(ui): address review findings from grumpy-review iteration 1 - Replace .expect() panics in TransferTokensScreen and ClaimTokensScreen constructors with graceful degradation via Option<QualifiedIdentity> and MessageBanner error display (PROJ-001 HIGH) - Fix CLAUDE.md referencing non-existent BannerHandleExt trait name, corrected to OptionBannerExt (PROJ-002 MEDIUM) - Update set_global to preserve message_type when same text appears with different severity (RUST-001 MEDIUM) - Standardize display_message to handle both Error and Warning across all 11 token screens (RUST-002 MEDIUM) - Replace 21 manual take().clear() patterns with take_and_clear() across 6 files (RUST-003 MEDIUM) - Remove unused OptionBannerExt::or_show_error method (RUST-004 MEDIUM) - Migrate update_token_config from old error_message pattern to set_error_banner closure (RUST-005 MEDIUM) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * perf(ui): replace per-frame QualifiedIdentity clone with borrow Use .as_ref() instead of .clone() in the ui() identity guard of TransferTokensScreen and ClaimTokensScreen. QualifiedIdentity (Identity + KeyStorage + BTreeMap + Vec) was being cloned 60x/sec; now only borrowed for display, with clones deferred to button-click paths that actually need ownership. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat(ui): add OptionResultExt::or_show_error for Option<T> Mirrors ResultBannerExt::or_show_error but for Option<T>: if None, displays a global error banner with the given message. Enables concise patterns like: identities.first().cloned().or_show_error(ctx, "No identities loaded") Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address review findings from grumpy-review iteration 2 - Standardize display_message to handle both Error and Warning across 13 non-token screens that were missed in iteration 1 (PROJ-001 MEDIUM) - Replace .expect() panic in AddKeyScreen::refresh() with graceful or_show_error() + unwrap_or_default() (PROJ-002 MEDIUM) - Rename OptionResultExt to OptionBannerShowExt to avoid confusion with ResultBannerExt (RUST-001 MEDIUM) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): handle Warning in add_new_identity_screen display_message Missed in the previous sweep — standardize display_message to handle both Error and Warning, matching all other screens. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): standardize display_message side-effect patterns across screens Guard side effects with Error|Warning match, use take_and_clear(), and remove redundant MessageBanner::set_global() call in 4 screens. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore(deps): update dashpay/platform to rev 570e3af0 Adapt to breaking changes in rust-dashcore (a05d256f → 2824e52a): - Replace removed FeeLevel enum with FeeRate::normal() direct calls - Replace removed WalletManager::create_unsigned_payment_transaction() with TransactionBuilder + WalletManager::get_change_address() - Replace removed DashSpvClientInterface/DashSpvClientCommand with direct Arc<SpvClient> for quorum lookups via get_quorum_at_height() - Replace removed start()+monitor_network() with client.run(token) - Add .await to now-async subscribe_sync/network/progress methods - Replace removed SyncState::Initializing with SyncState::WaitForEvents Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: log backtrace on panic * fix: panic in asset locks * fix(ui): address PR #604 review comments (CMT-001, CMT-002, CMT-003) - Fix QR scanner form reset matching wrong result type (CMT-001) - Remove dangerous identity fallback in token transfer screen (CMT-002) - Add fee-aware validation before credit transfers (CMT-003) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(spv): replace AsyncRwLock with ArcSwapOption for SPV client reference The SPV client reference only needs atomic set/clear (on start/stop) and wait-free reads (quorum lookups). ArcSwapOption is a better fit than AsyncRwLock<Option<Arc<...>>> — no lock contention, no async in blocking context, and consistent with how AppContext already uses ArcSwap for the SDK. Also fixes stale doc comment referencing removed DashSpvClientInterface. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address PR #604 review comments iteration 2 (transfer_tokens_screen) - Remove duplicate conflicting banner on missing identity (CMT-003) - Use generic banner messages with with_details() for errors (CMT-002) - Fix refresh to match specific token by contract+position (CMT-001) - Document error banner pattern in CLAUDE.md Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * doc: review guide updated * fix(ui): apply grumpy-review triage fixes — core infrastructure Address findings from code review triage (SEC-002, RUST-003, RUST-012, PROJ-005): preserve error context in payment tx builder, add INTENTIONAL comment for Debug-formatting accepted risk, remove dead import in SPV manager, fix CLAUDE.md error banner documentation, and add OptionBannerExt replace/replace_with_elapsed convenience methods. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): apply grumpy-review triage fixes — screen-level changes RUST-010: Add wallet_open_attempted guard to ~30 screens preventing per-frame try_open_wallet_no_password calls. Guard resets on wallet or identity selection change. PROJ-001: Migrate register/update contract and document action screens from BroadcastStatus::Broadcasting(u64) to refresh_banner with elapsed time display. RUST-001: Surface database errors via .or_show_error() in contacts_list instead of silently swallowing with let _ =. RUST-005: tokens_screen display_message only clears operation_banner on Error/Warning, preserving it on Success/Info. RUST-007: Remove dead Error(String, Instant) variant from TransitionBroadcastStatus in transition_visualizer_screen. RUST-009: Make identity Optional in claim_tokens_screen with degraded state handling when identity not found locally. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): handle OverallConnectionState::Error in connection banner Cover the new Error variant added by #650 (SPV sync error state) in the connection banner match arm. Shows an error banner directing users to the connection status tooltip for details. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: update Cargo.lock * build: reduce dependency debuginfo to fix <unknown> panic backtraces std's backtrace symbolizer (gimli) silently fails to resolve symbols when parsing ~790MB of DWARF debug sections in our 1.1GB debug binary, producing backtraces full of <unknown> frames. Set `debug = "line-tables-only"` for dependency crates while keeping full debuginfo for our own code. This cuts DWARF from 790MB to ~354MB and binary size from 1.1GB to 600MB, letting gimli actually resolve panic backtraces. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Pasta Lil Claw <pasta+claw@dashboost.org> Co-authored-by: PastaClaw <thepastaclaw@users.noreply.github.com> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Paul DeLucia <69597248+pauldelucia@users.noreply.github.com>
lklimek
added a commit
that referenced
this pull request
Mar 18, 2026
* docs: add unified message display design documents Add UX specification, technical architecture, and HTML mockup for the MessageBanner component that will replace the ~50 ad-hoc error/message display implementations across screens with a single reusable component. Key design decisions: - Per-screen MessageBanner with show()/set_message() API - All colors via DashColors (zero hardcoded Color32 values) - 4 severity levels: Error, Warning, Success, Info - Auto-dismiss for Success/Info (5s), persistent for Error/Warning - Follows Component Design Pattern conventions (private fields, builder, show) - No changes to BackendTask/TaskResult/AppState architecture Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add MessageType::Warning * chore: initial implementation * chore: docs * chore: self review * refactor(context): replace RwLock<Sdk> with ArcSwap<Sdk> (#600) * refactor(context): replace RwLock<Sdk> with ArcSwap<Sdk> Sdk is internally thread-safe (Arc, ArcSwapOption, atomics) and all methods take &self. The RwLock was adding unnecessary contention across backend tasks. Using ArcSwap instead of plain Sdk because reinit_core_client_and_sdk() needs to atomically swap the entire Sdk instance when config changes. ArcSwap provides lock-free reads with atomic swap for the rare write. Suggested-by: lklimek * fix: address CodeRabbit review findings for ArcSwap migration - Fix import ordering: move arc_swap::ArcSwap before crossbeam_channel - Remove redundant SDK loads in load_identity_from_wallet, register_dpns_name, and load_identity — use the sdk parameter already passed to these functions - Fix stale TODO referencing removed sdk.read().unwrap() API - Rename sdk_guard → sdk in transfer, withdraw_from_identity, and refresh_loaded_identities_dpns_names (no longer lock guards) - Pass &sdk to run_platform_info_task from dispatch site instead of reloading internally - Fix leftover sdk.write() call in context_provider.rs (RwLock remnant) - Add missing Color32 import in wallets dialogs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: address remaining CodeRabbit review feedback on ArcSwap migration - Move SDK load outside for loop in refresh_loaded_identities_dpns_names.rs so it's loaded once for the batch instead of on each iteration - Update stale TODO comment in default_platform_version() to reflect that this is a free function with no sdk access Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: consolidate double read-lock on spv_context_provider Clone the SPV provider in a single lock acquisition, then bind app context on the clone instead of locking twice. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: PastaClaw <thepastaclaw@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> * refactor: remove unused Insight API and show_in_ui config fields (#597) * refactor: remove unused Insight API references The `insight_api_url` field in `NetworkConfig` and its associated `insight_api_uri()` method were never used in production code (both marked `#[allow(dead_code)]`). Remove the field, method, config entries, env example lines, and related tests. https://claude.ai/code/session_01HWPmCJHT8KTZGP9bFiksjn * refactor: remove unused `show_in_ui` field from NetworkConfig The `show_in_ui` field was defined on `NetworkConfig` and serialized in `save()`, but never read by any production code to control network visibility. Remove the field, its serialization, env example lines, and test references. https://claude.ai/code/session_01HWPmCJHT8KTZGP9bFiksjn * fix: add missing `Color32` import in wallet dialogs https://claude.ai/code/session_01HWPmCJHT8KTZGP9bFiksjn --------- Co-authored-by: Claude <noreply@anthropic.com> * build: add Flatpak packaging and CI workflow (#589) * build: remove snap version * build: add Flatpak packaging and CI workflow Add Flatpak build manifest, desktop entry, AppStream metadata, and GitHub Actions workflow for building and distributing Flatpak bundles. Uses freedesktop 25.08 runtime with rust-stable and llvm21 extensions. No application source code changes required - works in SPV mode by default. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: address review findings for Flatpak packaging - Pin GitHub Actions to commit SHAs for supply chain security - Upgrade softprops/action-gh-release from v1 to v2.2.2 - Remove redundant --socket=x11 (fallback-x11 suffices) - Remove duplicate tag trigger preventing double builds on release - Remove duplicate env vars inherited from top-level build-options - Add Flatpak build artifacts to .gitignore - Add bugtracker URL to AppStream metainfo - Remove deprecated <categories> from metainfo (use .desktop instead) - Add Terminal=false and Keywords to desktop entry - Add disk space check after SDK install in CI - Rename artifact to include architecture suffix Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: simplify CI workflows for Linux-only releases - Remove "Free disk space" step from flatpak and release workflows - Remove Windows and macOS builds from release workflow - Use "ubuntu" runner image instead of pinned versions - Clean up unused matrix.ext references Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: attach to existing releases instead of creating new ones - Replace release-creating job with attach-to-release (only on release event) - Add 14-day retention for build artifacts - On tag push or workflow_dispatch, only upload artifacts (no release) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * revert: restore release.yml to original v1.0-dev version The release workflow changes were out of scope for the Flatpak PR. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: address CodeRabbit review comments - Fix CRLF line endings in Flatpak manifest (convert to LF) - Set app_id on ViewportBuilder to match desktop StartupWMClass - Use --locked flag for reproducible cargo builds in Flatpak - Rename --repo=repo to --repo=flatpak-repo to match .gitignore - Add architecture note for protoc x86_64 binary Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs: add Flatpak install instructions to README Add a dedicated section for installing via Flatpak on Linux, clarify that prerequisites are only needed for building from source, and rename "Installation" to "Build from Source" for clarity. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: match StartupWMClass to Flatpak app_id Use reverse-DNS format org.dash.DashEvoTool to match the Wayland app_id set via ViewportBuilder::with_app_id(). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: use ** glob for branch trigger to match feat/flatpak Single * doesn't match path separators in GitHub Actions branch filters. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add aarch64 Flatpak build and caching to CI - Add matrix strategy for parallel x86_64 and aarch64 builds - Patch protoc URL/sha256 per architecture at build time - Cache .flatpak-builder directory keyed on arch + manifest + lockfile - Pin actions/cache to SHA Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: convert desktop and metainfo files to LF line endings Flatpak builder validates desktop files and rejects CRLF line endings. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: cancel in-progress Flatpak builds on new push Add concurrency group keyed on git ref so a new push cancels any running build for the same branch or release. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: address PR review findings for Flatpak packaging - Remove unnecessary --filesystem=xdg-config/dash-evo-tool:create (Flatpak already redirects XDG_CONFIG_HOME to sandbox) - Add categories and keywords to AppStream metainfo for discoverability - Update README with both x86_64/aarch64 install commands, uninstall instructions, and Flatpak data path note - Clarify aarch64 comment in manifest to reference CI sed patching Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: workflow timeout and perms * fix: move permissions to job level in Flatpak workflow Step-level permissions are not valid in GitHub Actions. Move contents: write to the job level where it is needed for release attachment. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * build: cache Cargo registry and target in Flatpak CI Bind-mount host-side cargo-cache and cargo-target directories into the Flatpak build sandbox so CARGO_HOME and target/ persist across builds. Uses split restore/save with cleanup of incremental and registry/src (similar to Swatinem/rust-cache). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: scope cargo cache bind-mount sed to build-args only The previous sed matched --share=network in both finish-args and build-args, corrupting finish-args. Use a sed range to only target the build-args section. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Apply suggestions from code review --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> * chore: fix build * chore: use new error handling everywhere - not self reviewed * chore: use message banner to show progress * fix: start elapsed counter at 1s instead of 0s Aligns elapsed display with the countdown timer which already adds 1 to avoid showing "0s" immediately. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: rabbit review * Update src/app.rs Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> * chore: peer review * chore: fix build errors * fix(ui): make MessageBanner::set_global truly idempotent set_global() no longer resets timestamps, auto-dismiss timer, or logged flag when a banner with identical text already exists. This makes it safe to call every frame without log spam or timer restarts. Cherry-picked from origin/fix/spv-peer-timeout (08e3b3b). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address code review findings R01, R03, R06, R09, R10 - R01: Replace expect("No key selected") with graceful match + error banner in 11 token screens to prevent panics on missing signing key - R03: Remove dead backend_message field from AddExistingIdentityScreen - R06: Replace is_some() + unwrap() with idiomatic if-let-Some pattern in 10 token screens; use is_some_and() in structs.rs - R09: Add use imports for MessageBanner in 5 dashpay screens, replacing 22 fully-qualified crate::ui::components::MessageBanner:: calls - R10: Replace custom_dash_qt_error_message inline rendering with MessageBanner::set_global in network_chooser_screen Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address review findings SEC-08, SEC-09, RUST-015, SEC-05, SEC-07 - SEC-08: Restore safe if-let-Some pattern in WithdrawalScreen::refresh() to prevent double unwrap() panic on DB error or deleted identity - SEC-09: Restore original DB lookup in SendPaymentScreen::load_contact_info() replacing hardcoded "alice.dash" mock data - RUST-015: Revert unimplemented!() back to ui.label() in update_token_config MarketplaceTradeMode arm - SEC-05: Add success banners for contact request accept/reject in ContactRequests::display_task_result - SEC-07: Add MessageBanner::clear_all_global() and call it from AppState::change_network() to prevent stale banners leaking across network switches Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs: update coding conventions and message display guidance Add fallible constructor rule (Result<Self, ...> when they can fail), rename section to "General rules", and document MessageBanner idempotency (no guard needed for set_global). Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): replace expect/panic with graceful error handling (SEC-10) Replace all expect() calls in token screen constructors and confirmation handlers with MessageBanner error display. Constructors handle errors internally and return Self with degraded state, keeping create_screen() clean. refresh() methods now show errors via MessageBanner instead of tracing-only logging. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): accept impl Display/Debug in MessageBanner API Change MessageBanner public methods to accept `impl Display` for message text and `impl Debug` for details, instead of `&str`. Remove needless `&format!(...)` borrows across 27 call sites. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): remove error Failed to get best chain lock for mainnet, testnet, devnet, and local Fixes #633 * feat(ui): add automatic connection status banners Display persistent MessageBanner notifications based on network connection state transitions. Mode-aware messages guide users toward the right recovery action (RPC vs SPV). - Disconnected (RPC): "Disconnected — check that Dash Core is running" - Disconnected (SPV): "Disconnected — check your internet connection" - Syncing (RPC): "Syncing with Dash Core…" - Syncing (SPV): "SPV sync in progress…" - Synced: banner cleared Uses Option<OverallConnectionState> for change detection, with None as initial/post-network-switch sentinel to force first evaluation. Closes #667 (partial — action links deferred to follow-up) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): pass TaskError directly to with_details() to avoid double-formatting The previous code used `format!("{err:?}")` which produced a String, then `with_details()` applied `{:#?}` again — wrapping the output in quotes and escaping inner characters. Passing `&err` directly lets Debug format once. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): correct copy-paste error messages in token screens Replace "Burning" error messages that were copy-pasted from burn screen into freeze, destroy, and resume token screens with contextually correct messages. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): restore lost success/error messages in 5 screens Replace display_message() calls with MessageBanner::set_global() in screens where display_message() is now a side-effect-only handler and no longer displays messages directly. Affected screens: create_asset_lock_screen, wallets_screen (MineBlocks), address_table (export error), profile_screen (validation), contact_details. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): replace unwrap/expect with graceful error handling Replace double unwrap in transfer_screen refresh() with unwrap_or_else + MessageBanner error display, matching the pattern from withdraw_screen. SEC-002 tokens_screen skipped: the .expect() calls are only on compile-time embedded image data (include_bytes!) which is safe. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): migrate masternode_list_diff_screen to global MessageBanner Replace ~15 local ui_state.message assignments and custom render_message_banner() with MessageBanner::set_global() via the display_message() trait method. Remove the message field from UiState and the unused Color32 import. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): improve banner eviction logging and atomics - Upgrade BANNER_KEY_COUNTER from Relaxed to SeqCst ordering for future-proofing against multi-threaded usage - Log evicted banners at warn level in set_global() and replace_global() - Add comment explaining why show_global() always writes back Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: remove resolved TODO.md All items tracked in the unified message display TODO have been addressed or moved to the review findings. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(ui): add INTENTIONAL markers and API documentation - Document why with_details() accepts Debug (not Display): structured error context is more useful in diagnostic details panes - Document replace_global() fallback-to-add behavior as intentional - Add INTENTIONAL(SEC-003) marker for developer mode error details - Add INTENTIONAL(SEC-004) marker for BannerHandle Send+Sync safety Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): extract shared token validation and fix ordering - Add validate_signing_key() helper in tokens/mod.rs to eliminate duplicated signing key validation across 12 token screens - Move signing key validation BEFORE WaitingForResult state transition so users see immediate errors instead of loading spinner then error - Replace is_err()/unwrap() anti-pattern with idiomatic let-else blocks in freeze, mint, transfer, destroy_frozen_funds, unfreeze screens Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(ui): return Result from get_selected_wallet Replace &mut Option<String> error out-parameter with idiomatic Result<Option<Arc<RwLock<Wallet>>>, String>. Update 26+ callsites across identity, token, DashPay, and contract screens. Callsite patterns: unwrap_or_else with MessageBanner for user-visible errors, unwrap_or(None) where errors were previously silently ignored. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): resolve duplicate imports and clippy warnings Remove duplicate MessageBanner imports in create_asset_lock_screen and wallets_screen/mod. Fix needless_borrows_for_generic_args clippy lints in profile_screen, transfer_screen, and wallets_screen. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * style: reorder imports in masternode_list_diff_screen Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address review findings from MessageBanner migration audit Apply 13 triaged review fixes plus 1 bug fix across 22 files: - Remove dead error state fields (backend_message, error_message, Error variant) - Replace .expect() panics with graceful fallback + MessageBanner in token screens - Fix missing MessageBanner::show_global() on contracts documents screen - Migrate DocumentActionScreen inline errors to MessageBanner - Replace unwrap_or(None) with error-reporting fallback in DashPay screens - Fix replace_global idempotency and use relaxed atomic ordering in banner - Extract shared set_error_banner helper for 8 token screens - Restore correct Some(0) wallet index semantics - Document BannerHandle lifecycle in CLAUDE.md Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: validate token description length before sending to Platform (#530) * fix: validate token description length before sending to Platform Descriptions must be either empty or 3-100 characters long. Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com> * fix(ui): validate token description by char count, not byte length String::len() counts UTF-8 bytes, causing multi-byte characters (CJK, emoji) to be miscounted against the 3–100 limit. Switch to chars().count() and update all UI labels to surface the minimum. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.5 <noreply@anthropic.com> Co-authored-by: Lukasz Klimek <842586+lklimek@users.noreply.github.com> * refactor(ui): consolidate banner extension traits into message_banner Move BannerHandleExt and ResultBannerExt from banner_ext.rs into message_banner.rs where they belong. Merge take_and_clear() into OptionBannerExt trait (impl for Option<BannerHandle>) alongside or_show_error() for Option. Remove the separate banner_ext module and Clearable helper trait for simplicity. Apply review findings: DRY patterns (take_and_clear, or_show_error, load_identities_with_banner), fix .expect() panics in constructors, restore known_identities in refresh(), narrow pub field visibility, add ScreenLike doc comments, and update CLAUDE.md conventions. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * doc(tmp): review guide for pr 604 * fix(ui): address review findings from grumpy-review iteration 1 - Replace .expect() panics in TransferTokensScreen and ClaimTokensScreen constructors with graceful degradation via Option<QualifiedIdentity> and MessageBanner error display (PROJ-001 HIGH) - Fix CLAUDE.md referencing non-existent BannerHandleExt trait name, corrected to OptionBannerExt (PROJ-002 MEDIUM) - Update set_global to preserve message_type when same text appears with different severity (RUST-001 MEDIUM) - Standardize display_message to handle both Error and Warning across all 11 token screens (RUST-002 MEDIUM) - Replace 21 manual take().clear() patterns with take_and_clear() across 6 files (RUST-003 MEDIUM) - Remove unused OptionBannerExt::or_show_error method (RUST-004 MEDIUM) - Migrate update_token_config from old error_message pattern to set_error_banner closure (RUST-005 MEDIUM) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * perf(ui): replace per-frame QualifiedIdentity clone with borrow Use .as_ref() instead of .clone() in the ui() identity guard of TransferTokensScreen and ClaimTokensScreen. QualifiedIdentity (Identity + KeyStorage + BTreeMap + Vec) was being cloned 60x/sec; now only borrowed for display, with clones deferred to button-click paths that actually need ownership. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat(ui): add OptionResultExt::or_show_error for Option<T> Mirrors ResultBannerExt::or_show_error but for Option<T>: if None, displays a global error banner with the given message. Enables concise patterns like: identities.first().cloned().or_show_error(ctx, "No identities loaded") Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address review findings from grumpy-review iteration 2 - Standardize display_message to handle both Error and Warning across 13 non-token screens that were missed in iteration 1 (PROJ-001 MEDIUM) - Replace .expect() panic in AddKeyScreen::refresh() with graceful or_show_error() + unwrap_or_default() (PROJ-002 MEDIUM) - Rename OptionResultExt to OptionBannerShowExt to avoid confusion with ResultBannerExt (RUST-001 MEDIUM) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): handle Warning in add_new_identity_screen display_message Missed in the previous sweep — standardize display_message to handle both Error and Warning, matching all other screens. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): standardize display_message side-effect patterns across screens Guard side effects with Error|Warning match, use take_and_clear(), and remove redundant MessageBanner::set_global() call in 4 screens. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore(deps): update dashpay/platform to rev 570e3af0 Adapt to breaking changes in rust-dashcore (a05d256f → 2824e52a): - Replace removed FeeLevel enum with FeeRate::normal() direct calls - Replace removed WalletManager::create_unsigned_payment_transaction() with TransactionBuilder + WalletManager::get_change_address() - Replace removed DashSpvClientInterface/DashSpvClientCommand with direct Arc<SpvClient> for quorum lookups via get_quorum_at_height() - Replace removed start()+monitor_network() with client.run(token) - Add .await to now-async subscribe_sync/network/progress methods - Replace removed SyncState::Initializing with SyncState::WaitForEvents Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * chore: log backtrace on panic * fix: panic in asset locks * fix(ui): address PR #604 review comments (CMT-001, CMT-002, CMT-003) - Fix QR scanner form reset matching wrong result type (CMT-001) - Remove dangerous identity fallback in token transfer screen (CMT-002) - Add fee-aware validation before credit transfers (CMT-003) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor(spv): replace AsyncRwLock with ArcSwapOption for SPV client reference The SPV client reference only needs atomic set/clear (on start/stop) and wait-free reads (quorum lookups). ArcSwapOption is a better fit than AsyncRwLock<Option<Arc<...>>> — no lock contention, no async in blocking context, and consistent with how AppContext already uses ArcSwap for the SDK. Also fixes stale doc comment referencing removed DashSpvClientInterface. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(ui): address PR #604 review comments iteration 2 (transfer_tokens_screen) - Remove duplicate conflicting banner on missing identity (CMT-003) - Use generic banner messages with with_details() for errors (CMT-002) - Fix refresh to match specific token by contract+position (CMT-001) - Document error banner pattern in CLAUDE.md Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * [claudesquad] update from 'testing research' on 02 Mar 26 13:25 CET (paused) * test: add backend E2E test harness and SPV wallet test Add test-only accessors (db(), wallets()) on AppContext gated behind cfg(test/testing), fix compilation errors in the backend-e2e test (private field access, unused import), and apply nightly rustfmt. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * feat: add DASH_EVO_DATA_DIR env var to override app data directory Allows tests and CI to redirect all app data (database, SPV chain state, .env config) to a temp directory. The backend-e2e test harness now uses this to achieve full data isolation. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * test: add backend E2E test framework with shared state and funded wallets Evolve the prototype backend E2E test harness into a reusable framework: - LazyLock shared BackendTestContext with persistent workdir, SPV, and framework wallet (funded via E2E_WALLET_MNEMONIC or testnet faucet) - Task runner wrapper, polling wait helpers, faucet HTTP client - Identity key derivation helpers for wallet-funded registration - Six test scenarios: SPV wallet, identity create, identity withdraw, send/receive funds, fetch contracts, register DPNS name - Move default_identity_key_specs() from UI to backend_task::identity (domain logic, not UI concern) and make IdentityKeys fields pub - Add dashpay_contract_id() test accessor to AppContext Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(test): use tokio::sync::OnceCell instead of LazyLock for async init LazyLock triggers synchronously inside the #[tokio::test] runtime, causing "Cannot start a runtime from within a runtime" when the init function calls block_on(). Switch to tokio::sync::OnceCell with an async init() method so shared state initialization runs cooperatively within the existing tokio runtime. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(test): handle persistent DB and SPV balance sync in E2E harness - Wait for SPV to sync existing wallet balance before checking if faucet funding is needed (pre-funded wallets need time to discover on-chain UTXOs) - Handle "already imported" error gracefully when framework wallet exists in persistent DB from a previous run Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(test): wait for spendable balance and retry sends in E2E harness The SPV wallet reports total balance (including unconfirmed) but only confirmed/IS-locked UTXOs are available for transaction building. This caused "Insufficient funds" errors when tests tried to spend immediately after receiving funds. - Add wait_for_spendable_balance() that checks confirmed_balance_duffs() and triggers reconcile_spv_wallets() on each poll iteration - Add retry logic (5 attempts, 10s backoff) to create_funded_test_wallet() for sends that fail with InsufficientFunds - Wait for framework wallet change output to become spendable after each send so subsequent calls don't fail - Add wait_for_spendable_balance() before identity registration in all identity/DPNS tests - Add send_with_retry() helper in send_funds test - Add developer-facing README.md for the test framework Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(test): use tokio-shared-rt for shared runtime and sweep orphaned wallets Replace per-test tokio runtimes with tokio-shared-rt's global shared runtime to prevent SPV background tasks from dying between test functions. Add automatic orphaned wallet fund recovery during setup — wallets persist in DB, so on next run the harness sweeps funds back to the framework wallet. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: handle OverallConnectionState::Error in connection banner Semantic merge conflict from v1.0-dev: PR #650 added an Error variant to OverallConnectionState, which our connection banner match didn't cover. Show an error banner when SPV sync enters the error state. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * docs(test): improve backend-e2e documentation and funding UX - Add backend E2E section to CLAUDE.md pointing to the full README - Document .env file handling (project root vs workdir) and precedence - Fix test attribute in README: tokio::test → tokio_shared_rt::test(shared) - Update init sequence to reflect current code (spendable wait, orphan sweep) - Document automatic cleanup-on-init of orphaned test wallets - Raise minimum balance threshold from 1 to 10 tDASH - Always panic with receive address when faucet fails and balance is below minimum (previously only panicked on zero balance) - Show receive address in spendable-balance timeout warning Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: patch SPV UTXO spendability flags before coin selection Upstream key-wallet-manager (rust-dashcore) never sets is_confirmed or is_instantlocked on UTXOs, but CoinSelector requires one of them. This caused "No UTXOs available for selection" errors despite having balance. Workaround infers status from block inclusion (height > 0 → confirmed, height == 0 → IS-locked). Ref: dashpay/rust-dashcore#514 Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * test: harden backend E2E tests with retry logic and calibrated amounts - Increase funding amounts to avoid insufficient-funds flakes - Add 3-attempt retry for identity registration (chain height sync) - Retry on "No UTXOs" in send_with_retry alongside "Insufficient" - Wait for spendable balance in create_funded_test_wallet before return - Add CI workflow for backend E2E tests Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Update test command in backend E2E workflow * ci: merge backend-e2e workflow into tests workflow (#727) * Initial plan * ci: merge backend-e2e workflow into tests workflow as an additional step Co-authored-by: lklimek <842586+lklimek@users.noreply.github.com> * test: add cleanup_only noop test to sweep orphaned wallets Add a standalone test that triggers BackendTestContext initialization, which runs cleanup_test_wallets() as its final step. This can be run as a dedicated CI step after the E2E suite to sweep orphaned wallet funds back to the framework wallet. Run with: cargo test --test backend-e2e --all-features -- --ignored --nocapture cleanup_only Co-authored-by: lklimek <lklimek@users.noreply.github.com> Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * Add cleanup step for E2E test wallets Added a cleanup step for E2E test wallets in the workflow. * Simplify E2E test workflow conditions Removed conditional checks for E2E_WALLET_MNEMONIC in test steps. --------- Co-authored-by: copilot-swe-agent[bot] <198982749+Copilot@users.noreply.github.com> Co-authored-by: lklimek <842586+lklimek@users.noreply.github.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: lklimek <lklimek@users.noreply.github.com> Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> * fix: address PR #673 review comments from triage Production code: - Extract patch_utxo_spendability_flags() helper to deduplicate workaround in estimate_fallback_amount and build_unsigned_payment_tx - Add IdentityKeys::new() constructor - Make wallet+address persistence atomic via store_wallet_with_addresses() - Add pending_wallet_selection after wallet creation Test code: - Add DPNS registration retry for identity propagation delay - Use u64 hex for DPNS name uniqueness (CMT-023) - Calibrate test funding amounts per reviewer feedback - Add fragility note on string-match wallet detection (CMT-006) - Add TODO for identity fund withdrawal in cleanup (CMT-032) - Add INTENTIONAL comment for bounded channel design (CMT-017) - Add balance assertion for return leg in send_funds test - Log reconcile_spv_wallets errors in wait helpers - Use is_ok_and for clarity in spv_wallet test Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: remove UTXO spendability workaround, update platform dep Remove `patch_utxo_spendability_flags()` that faked IS-locked status on mempool UTXOs. Wait for upstream fix (dashpay/rust-dashcore#514) to properly set is_confirmed/is_instantlocked flags on UTXOs. Also: - Update dashpay/platform rev to aa86b74f7e2 - Adapt to upstream API: FeeLevel→FeeRate, remove NetworkExt import - Fix retry to catch "No UTXOs" errors in test harness Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: deduplicate default_identity_key_specs Move the single canonical copy to backend_task::identity::mod and have the UI screen import it, eliminating ~240 lines of duplicated function and tests. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * ci: temporarily disable backend E2E tests in CI The backend E2E tests need updates after the TaskError migration (#739) changed AppContext field visibility. Commenting out the CI steps until the tests are adapted. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: type-safe wallet creation and identity key visibility - Restrict IdentityKeys fields to pub(crate) to prevent private key exposure outside the crate - Change register_wallet() to return TaskError instead of String, using proper rusqlite error matching via is_unique_constraint_violation() and a new WalletAlreadyImported variant - Change Wallet::new_from_seed() to accept Option<&Secret> for password instead of Option<&str>, keeping sensitive data in the Secret wrapper - Change Wallet::new_from_seed() to return TaskError instead of String, with a new WalletKeyDerivationFailed variant for derivation errors - Move build_identity_registration() and get_receive_address() from test helpers to production code in src/backend_task/identity/mod.rs - Extract is_unique_constraint_violation() to src/database/mod.rs as a shared pub(crate) utility, removing the duplicate in import_mnemonic_screen - Update all callers in add_new_wallet_screen and import_mnemonic_screen Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * test: address PR #673 review comments for E2E test framework - Move framework modules (harness, task_runner, wait, funding, cleanup, identity_helpers) into tests/backend-e2e/framework/ subdirectory - Make E2E_WALLET_MNEMONIC required (panic with instructions if unset) - Remove auto-faucet from initialization flow (keep as helper) - Remove retry loops in identity_create and identity_withdraw tests - Remove unnecessary wait_for_spendable_balance calls (already done by create_funded_test_wallet) - Replace all println!/eprintln! with tracing macros - Initialize tracing subscriber in harness init - Add "No spendable funds" and "spendable" to send retry conditions - Remove stale "other agent" NOTE comments from identity_helpers - Consolidate funding logic (harness delegates to funding module) - Update README for required mnemonic and new directory structure Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: reconcile production and test code after type-safe refactor Adapt test framework to production API changes: - Use IdentityKeys::new() constructor (fields now pub(crate)) - Match TaskError::WalletAlreadyImported variant instead of string - Allow dead_code on faucet helpers (kept but not auto-called) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * refactor: thread data_dir as explicit parameter to eliminate env var dependency Add `data_dir: PathBuf` field to `AppContext` and thread it through `Config::load_from()`, `Config::save()`, `SpvManager::new()`, `start_dash_qt()`, and `create_dash_core_config_if_not_exists()`. This enables E2E tests to specify their data directory without mutating process-wide environment variables, making parallel test execution safe. The `DASH_EVO_DATA_DIR` env var is still checked in production via `app_user_data_dir_path()`, but the resolved path is threaded through as a value rather than re-read from env on every call. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix(test): simplify funded wallet creation and make spendable balance check reliable - Remove retry loop from create_funded_test_wallet; wait for full amount_duffs instead of 1 duff in spendable balance check - Add Wallet::spv_confirmed_balance() that returns None when SPV hasn't synced yet (no max_balance fallback) - Use spv_confirmed_balance() in wait_for_spendable_balance so the wait never gets false positives from the fallback - Remove --test-threads=1 requirement from README (unsafe set_var was the only reason; data_dir is now threaded explicitly) Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> * fix: remove send retry loop, handle PRIMARY KEY constraints, isolate SPV test dirs - Remove send_with_retry() from send_funds.rs; use wait_for_spendable_balance before each send instead - Add SQLITE_CONSTRAINT_PRIMARYKEY (1555) to uniqueness check alongside SQLITE_CONSTRAINT_UNIQUE (2067) - Use tempfile::TempDir in SPV tests instead of fixed /tmp/spv-test path to prevent state leaks and support concurrent test runs Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com> --------- Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com> Co-authored-by: Pasta Lil Claw <pasta+claw@dashboost.org> Co-authored-by: PastaClaw <thepastaclaw@users.noreply.github.com> Co-authored-by: Copilot <175728472+Copilot@users.noreply.github.com> Co-authored-by: Paul DeLucia <69597248+pauldelucia@users.noreply.github.com> Co-authored-by: Copilot <198982749+Copilot@users.noreply.github.com> Co-authored-by: claude[bot] <41898282+claude[bot]@users.noreply.github.com> Co-authored-by: lklimek <lklimek@users.noreply.github.com>
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Issue being fixed or feature implemented
SPV sync gets permanently stuck in "Syncing" status when a fatal error occurs
(e.g., masternode QRInfo failure). The connectivity icon stays orange instead
of turning red, and no error message is shown to the user.
Symptoms observed in logs:
Despite these errors, the UI showed no indication of failure.
Root cause: Two gaps in error detection:
spawn_sync_event_handlerignoredSyncEvent::ManagerErrorevents entirely.spawn_progress_watcheronly checkedis_synced()and defaulted everythingelse to
SpvStatus::Syncing, never checking forSyncState::Error.Additionally, an upstream bug in dash-spv (dashpay/rust-dashcore#469) means
the progress channel never receives the error state update, making gap #2
a dead path until the library is fixed.
What was done?
Error detection (manager.rs)
Added two error detection paths in
src/spv/manager.rs:spawn_sync_event_handler(primary path): Now handlesSyncEvent::ManagerError— transitionsSpvStatustoErrorand storesthe error message in
last_error. This works reliably because dash-spvalways emits this event on manager failure.
spawn_progress_watcher(defense-in-depth): Now checkswatch_progress.state() == SyncState::Errorand transitions accordingly.Currently blocked by upstream bug but will activate once
bug: SyncManager does not emit progress update on manager error path rust-dashcore#469 is fixed.
Distinct Error UI state (connection_status.rs, top_panel.rs, theme.rs)
Previously,
SpvStatus::Errormapped toOverallConnectionState::Disconnected—the same red static circle as "never connected". Users couldn't distinguish
between the two. Now:
OverallConnectionState::Errorvariant (repr3)SpvStatus::Errormaps toOverallConnectionState::Errorexplicitlyfrom
spv_last_error(populated fromSpvManager::status().last_error)DashColors::sync_error_color()for the magenta-red indicator colorErrorarm ("Connection error")Related upstream issues
SyncManagerdoes not emit progress update onerror path
robust
How has this been tested?
cargo clippy --all-features --all-targets -- -D warnings— cleancargo +nightly fmt --all— cleandocs/ai-design/2026-02-24-spv-sync-error-status/manual-test-scenarios.mdBreaking Changes
None
🤖 Co-authored by Claudius the Magnificent AI Agent
Summary by CodeRabbit
New Features
Documentation