feat(desktop): port tray/Tauri integration onto the client/server settings model - #216
Merged
Merged
Conversation
This was referenced Aug 30, 2026
![github-advanced-security[bot]](https://avatars.githubusercontent.com/in/57789?s=60&v=4){kind=small}
Comment on lines
+30
to
+34
| impl Default for FoundryInstallActionArgs { | ||
| fn default() -> Self { | ||
| Self { approved: false } | ||
| } | ||
| } |
Comment on lines
+30
to
+34
| impl Default for FoundryInstallActionArgs { | ||
| fn default() -> Self { | ||
| Self { approved: false } | ||
| } | ||
| } |
| }; | ||
| let json = serde_json::to_string(&status).unwrap(); | ||
| let back: FoundryLocalStatus = serde_json::from_str(&json).unwrap(); | ||
| assert_eq!(back.cli_found, true); |
| let json = serde_json::to_string(&status).unwrap(); | ||
| let back: FoundryLocalStatus = serde_json::from_str(&json).unwrap(); | ||
| assert_eq!(back.cli_found, true); | ||
| assert_eq!(back.service_running, false); |
This was referenced Aug 30, 2026
…erShell; retire host-tray.rs - NotificationBackend::PowerShell renamed to Native (serde alias keeps reading settings persisted under the old name) — the actual toast delivery is being replaced with native windows-rs calls in ledgerr-host, this crate only owns the shared data types. - SettingsStore::with_backend(path, backend) added alongside new(), so tests can construct an isolated JsonFileBackend directly instead of going through Windows registry auto-selection — new()'s registry backend ignores its path argument entirely and always targets one fixed production key, which is correct for real callers but means any test using new() over a tempdir shares that one global mutable key with every other test and any real running instance. - crates/ledgerr-host/src/bin/host-tray.rs moved to src/bin/legacy/ as reference-only (git mv staged this alongside the above; it's the first step of retiring the standalone host-tray.exe binary — completed in the following commits, which remove its [[bin]] entry and repoint host-tauri's tray at the same shared runtime it used).
…le host-tray refs settings_server.rs's tests hit the same registry-auto-selection issue fixed in ledgrrr-settings — repoint them at SettingsStore::with_backend + JsonFileBackend explicitly. status.rs/service_control.rs still listed host-tray as a tray-binary candidate; it's being retired in favor of host-tauri.exe, which already sorts first in that list.
…ndows-rs toasts The tray's toast notifications shelled out to powershell.exe running Import-Module BurntToast; New-BurntToastNotification — requiring a separately-installed PowerShell module on every machine. A native windows::UI::Notifications-based ToastNotifier already existed in this crate (notification::windows_toast) but was never wired into the tray's actual notification path — only its own unused tests exercised it. Added notify::native::NativeToastNotifier, implementing the same notify::types::Notifier trait the tray already consumes, wrapping that existing native implementation (and the stderr fallback on non-Windows) instead of shelling out to any external process.
…ubmenu native.rs: drop the unused push_check helper, DYNAMIC_TEXT_IDS and CHECK_ITEM_IDS constants, and fix a few compiler-flagged unused-mut/ unused-must-use warnings. Group the four notification-event toggles (approval/submitted/failed/ completed) into a "Notify me for" submenu instead of sitting flat among the other 10 top-level menu items — cuts the flat menu from 14 items to 11 and keeps the related settings together.
… down settings service Ports this session's earlier work (originally built against a SettingsStore this branch predates) onto the client/server settings architecture already on main (SettingsClient, ledgrrr-service): - TrayCommand::ShowWindow's action is now an injected closure instead of a hardcoded spawn of a separate process, so the same run()/ handle_command can be reused by host-tauri (which shows its own webview) without any dependency on a standalone binary. - Collapsed 7 near-identical toggle handlers into negate()/ apply_toggle() helpers. - Any single command's error (most realistically: ledgrrr-service isn't running yet) is now logged and the loop continues, instead of propagating via `?` out of run() and killing the whole tray — and, via host-tauri's exit-on-return, the whole app — over a transient failure. Only an explicit Quit ends the loop. - run()'s own initial settings load falls back to defaults on failure instead of erroring out before ever creating the tray icon at all — confirmed live: with no settings service running, the tray now still appears (previously it silently never initialized). - NotificationBackend::PowerShell -> Native throughout.
…ost-tray.exe Replaces the minimal Show/Exit-only stub with the shared tray::runtime::run(), injecting "show/unminimize/focus the Tauri webview" as the window-show action. host-tauri now has the full toggle/notification/toast menu that was previously only in the standalone host-tray.exe, fixing the toast-notifications-never-wired-up bug on the main app's tray. Also: main.rs's window-visibility check and enable_tray check both silently fell back to unhelpful defaults (or, for enable_tray, silently skipped tray setup entirely) whenever the settings load failed — exactly the case when ledgrrr-service isn't running yet. Both now log and use an explicit, sensible default (visible; tray enabled) instead.
Removes its [[bin]] entry now that host-tauri.exe has full feature parity (previous commit). Source stays at src/bin/legacy/host-tray.rs for reference; it no longer compiles as part of any normal build.
host-tray no longer exists as a build target (previous commit).
PR #187 (spike/sysml-v2-parser-roundtrip) found holon-viz's SysmlV2Emitter produced invalid SysML v2 (comment-swallowed closing brace, `block def` instead of `part def`). Both bugs were independently fixed on main in PR #183 (c106a5e) before this doc landed. Cherry-picks only the investigation doc; the throwaway sysml-v2-parser-spike crate stays on the closed PR's branch since its regression test now asserts stale (pre-fix) behavior.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com> Claude-Session: https://claude.ai/code/session_015kUr7Kf9wN15TPqDiKsiKw
…, not this repo's own MSIX)
elasticdotventures
force-pushed
the
feat/tray-client-server-port
branch
from
August 30, 2026 09:00
1176a46 to
1cac1be
Compare
This PR's own with_backend()/JsonFileBackend fix for the registry-leak issue (SettingsStore::new's Windows registry backend ignores its path arg and always targets one fixed production key) landed in ledgrrr-settings and ledgerr-desktop-agent, but missed 3 test files in ledgerr-host itself — the crate that actually owns the tray's real SettingsStore usage: - tests/settings_roundtrip.rs (5 tests) - tests/settings_atomicity.rs (2 tests) - tests/tray_wiring_smoke.rs (tray_enable_roundtrips_through_store) On a real Windows box these all still called SettingsStore::new(tempdir) directly, silently opening the one production registry key instead of an isolated per-test backend — the exact corruption/cross-contamination bug this PR claims to have fixed elsewhere, and in the worst case (tray_enable_roundtrips_through_store) capable of writing enable_tray: false into a real running host-tauri.exe's live settings from `cargo test`. CI never caught it: the only Windows job runs a packaging smoke test, not `cargo test`, and on Linux the registry backend is compiled out entirely so these pass cleanly via the (correctly path-scoped) JsonFileBackend fallback regardless. Applied the same with_backend(path, JsonFileBackend::new(path)) pattern already used in ledgrrr-settings/src/store.rs and ledgerr-desktop-agent/src/settings_server.rs. Verified: cargo test -p ledgerr-host --tests — 112 passed, 0 failed.
Member
Author
|
Reviewed and pushed a fix directly to this branch (commit a10e402) — not merging yet, pending CI on the fixed commit and your own pass over the follow-up notes below. Fixed: this PR's own registry-key test-isolation fix ( Separately: this branch was also stale (based on a pre-Containerfile-fix Confirmed working, no changes needed:
Not fixed — worth a follow-up issue, not a blocker:
Please re-check CI on the new head commit before merging. |
- Restore docs/superpowers/specs/2026-08-29-tray-tauri-integration-design.md and its companion plan doc from the superseded feat/windows-desktop-dogfood branch — Justfile referenced the spec doc, but it was never ported when this branch superseded that one. Retrieved verbatim, status line updated to note the branch handoff. - Add invoke_reports_blocked_on_non_windows_even_when_approved, gated #[cfg(not(windows))] — exercises the untested !plan.executable_now branch in foundry_install_plan.rs::invoke() on non-Windows CI targets. Deliberately excluded from Windows builds: on Windows, executable_now is always true, so calling invoke(approved: true) there would reach the real winget-spawning code path as a side effect of running the test suite — the existing invoke_without_approval_never_launches test already covers the approval gate safely on every target.
…-store + foundry_install_plan conflicts - crates/ledgrrr-settings/src/store.rs: this branch and #234 (already merged) independently fixed the same bug (SettingsStore::new's Windows registry backend ignoring its path argument in tests) with different APIs -- this branch's with_backend(path, backend) and #234's new_json_file(path). Kept both: with_backend stays the general primitive (used as-is by 4 other test call sites in this branch), new_json_file is now a thin wrapper around it, so neither PR's capability is lost. - crates/ledgerr-desktop-agent/src/foundry_install_plan.rs: add/add conflict from divergent history (this branch and #235, also merged, each added the file); both sides already had identical content including the same non-Windows-only test, so no real merge was needed.
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
3 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.
Summary
Supersedes PR #213 (closed in favor of this one — #213 was built against a pre-client/server
SettingsStoresnapshot ofmainand doesn't apply cleanly on top of the settings-service refactor that landed independently while that work was in progress).Same functional goal as #213 —
host-tauri.exe's Windows tray previously had no toast/notification support at all (a minimal Show/Exit-only stub), while a separate standalonehost-tray.exehad the full feature set but no integration with the main app. This ports that consolidation onto theSettingsClient/ledgrrr-servicearchitecture already onmain.host-tauri.exe's Windows tray through the same shared, testedtray::runtime::run()the standalone tray used, now taking aSettingsClient— gives it the full toggle/notification/toast menu, fixing the reported "toast notifications don't work" bugwindows-rstoast implementation (notify::native::NativeToastNotifier, wrapping the already-present but previously-unwirednotification::windows_toast::ToastNotifier) — no external process, no module install required.NotificationBackend::PowerShellrenamed toNative(serde alias preserves settings persisted under the old name)host-tray.exeas a build target (source kept undersrc/bin/legacy/for reference;host-window.exe/Slint were already retired onmainvia refactor(ledgerr-host): remove legacy Slint host #174)SW_SHOWvsSW_RESTORE)ledgrrr-servicenot running (the realistic first-launch state), the app previously either defaulted the main window to hidden/inert states or — worse — silently skipped tray setup entirely, so no tray icon ever appeared. Bothmain.rs's window-visibility check and itsenable_traycheck now log and fall back to a sensible default (visible; tray enabled) instead. Verified live: tray icon now appears with the settings service fully down.ledgrrr-settings,ledgerr-desktop-agent):SettingsStore::new()'s Windows registry backend ignores itspathargument and always targets one fixed production key — correct for real callers, but meant every test using a tempdir path shared that one global mutable key. AddedSettingsStore::with_backend()so tests can inject an isolatedJsonFileBackendexplicitly.Test plan
cargo test -p ledgerr-host -p ledgrrr-settings --lib --tests --bins— 95 passing. 2 known pre-existing failures inchat_tools(Windows-path-in-TOML escaping bug tied to this machine's username) confirmed to fail identically on a clean, unmodifiedmain— not introduced by this branch.host-tauri.exewithledgrrr-servicenot running: main window appears (visible-by-default fallback), tray icon appears (L3dg3rrTrayWindowconfirmed via window enumeration) with the full menu, not just Show/ExitBurntToastmodule confirmed not installed on the validation machine;notify-test.exe --backend nativereturns a real"status": "ready"/"message": "toast sent"result via the nativewindows::UI::NotificationsAPI🤖 Generated with a build+review process (implementation, live validation against a running host-tauri.exe, iterative fixes based on what that surfaced).
https://claude.ai/code/session_015kUr7Kf9wN15TPqDiKsiKw