Advance app runtime ownership and modernization docs

AGENTS.md

@@ -30,11 +30,12 @@ Read these first:
 - Commit and push each verified successful progress slice.
 - Prefer larger coherent slices over tiny checkpoints, but keep docs/debt updated
   with each slice.
-- After a verified slice is committed and pushed, reset conversation context
+- Treat automatic compaction as a failure mode to avoid. Keep active context
-  before starting the next slice when practical, especially if the thread is
+  small, commit and push before the thread grows large, and reset conversation
-  approaching automatic compaction. Record all needed resume state in committed
+  context between verified slices when practical instead of carrying excess
-  code/docs first so the next thread can restart from `AGENTS.md`, roadmap/debt,
+  history forward. Record all needed resume state in committed code/docs first
-  and git history instead of relying on chat transcript context.
+  so the next thread can restart from `AGENTS.md`, roadmap/debt, and git
+  history instead of relying on chat transcript context.
 - Do not revert user changes. Unrelated untracked notes, such as
   `docs/human-review-notes.md`, should be left alone unless explicitly requested.
 - Use CMake as the source of truth. Legacy Visual Studio project files are not the
@@ -42,8 +43,9 @@ Read these first:
 - Use `apply_patch` for manual source/doc edits.
 - For delegated work, follow `docs/modernization/director-workflow.md`: the
   coordinator keeps integration locally, assigns direct worker tasks, uses
-  `gpt-5.3-codex-spark` workers by default, and gives them the exact project
+  `gpt-5.4-mini` workers by default, and gives them a minimal task packet with
-  context needed so they do not spend tokens re-reading repo docs.
+  only the build, test, and code-exploration context needed so they do not
+  spend tokens re-reading repo docs.
 
 ## Build And Test
 

cmake/PanoPainterSources.cmake

@@ -86,6 +86,7 @@ set(PP_PLATFORM_LINUX_SOURCES
 
 set(PP_PANOPAINTER_APP_SOURCES
     src/app.cpp
+    src/app_runtime.cpp
     src/app_cloud.cpp
     src/app_commands.cpp
     src/app_dialogs.cpp

docs/modernization/capability-map.md

@@ -1,7 +1,7 @@
 # PanoPainter Capability Map
 
 Status: live
-Last updated: 2026-06-05
+Last updated: 2026-06-16
 
 This map is the preservation checklist for the modernization. When a component
 is extracted, update the relevant rows with the owning component, test label,
@@ -59,6 +59,7 @@ and validation command.
 | Yoga layout | `Node` | `pp_ui_core` | Deterministic geometry fixtures |
 | Generic controls | `NodeButton`, sliders, text, images | `pp_ui_core` | Event dispatch, layout, ownership-handle, callback-disconnect, and destroy-during-callback tests |
 | PanoPainter panels/dialogs | `NodePanel*`, `NodeDialog*` | `pp_panopainter_ui`, `pp_ui_core` | UI automation scripts, command-dispatch view models, pure overlay lifetime tests, retained overlay-adapter build coverage, retained popup/dialog lifetime tests |
+| UI ownership and thread affinity | `Node`, `LayoutManager`, `App` UI queue, retained callbacks | `pp_ui_core`, app runtime service, `pp_panopainter_ui` | Checked-handle dispatch, scoped callback disconnect, destroy-during-callback, close-during-dispatch, and UI-thread post/drain/shutdown coverage |
 | Canvas viewport UI | `NodeCanvas` | `pp_panopainter_ui`, `pp_paint_renderer` | Input-to-command automation |
 | Settings UI | `Settings`, `NodeSettings` | `pp_assets`, `pp_panopainter_ui` | Round-trip settings |
 
@@ -67,6 +68,7 @@ and validation command.
 | Capability | Current Area | Target Owner | Required Tests |
 | --- | --- | --- | --- |
 | Mouse/keyboard/touch/gestures/cursor | `App`, platform entrypoints | `pp_app_core`, `pp_platform_api`, `pp_platform_*`, app | Cursor visibility decision tests, platform service dispatch tests, synthetic event playback |
+| Render/UI task dispatch and worker shutdown | `App`, `Canvas`, retained worker threads, platform entrypoints | app runtime service, `pp_foundation`, `pp_platform_*` | Render/UI queue order, same-thread dispatch, cancellation, shutdown drain, and no-detached-worker ownership coverage |
 | Wacom pressure | `WacomTablet` | `pp_platform_windows` | Adapter smoke with fallback |
 | Clipboard/file picker/share/display | `App` platform methods | `pp_app_core`, `pp_platform_api`, `pp_platform_*` | Clipboard read/write, share saved-path, picked-path, and display-file decision tests, platform service display/share/picker dispatch tests, platform smoke or mocked service |
 | Virtual keyboard | `App`, platform entrypoints | `pp_app_core`, `pp_platform_api`, `pp_platform_*` | Keyboard visibility decision tests, platform service dispatch tests, platform smoke |

docs/modernization/debt.md

@@ -335,6 +335,10 @@ agent or engineer to remove them without reconstructing context from chat.
   routes the stroke preset/brush/dual-brush/pattern popup open-close flow through
   checked overlay handles and explicit partial-open cleanup; migration is still
   pending in remaining panel families.
+- 2026-06-16: `DEBT-0063` was narrowed again. `src/node_panel_stroke.cpp` now
+  routes stroke popup close/reset through a shared overlay helper, and
+  `src/node_popup_menu.cpp` now keeps the popup alive through selection dispatch
+  with scoped ownership instead of raw `this` callbacks.
 - 2026-06-14: `DEBT-0063` was narrowed again. `src/node_panel_brush.cpp` now
   routes brush popup-menu open-close flow through checked overlay handles and
   handle-based close on selection.

docs/modernization/director-workflow.md

@@ -1,7 +1,7 @@
 # Modernization Coordinator Workflow
 
 Status: live
-Last updated: 2026-06-14
+Last updated: 2026-06-16
 
 Use this workflow when the user explicitly asks for subagents, delegation, a
 coordinator, or parallel agent work. Do not spawn subagents just because a task
@@ -16,10 +16,16 @@ layer.
 
 - Save main-thread tokens by keeping implementation and focused lookup out of
   the coordinator context.
+- Treat thread compaction as wasted budget. Prefer small active context,
+  committed resume state, and fresh follow-on threads over carrying long stale
+  history.
 - Keep each implementation slice measurable, validated, committed, and pushed.
 - Avoid merge conflicts by giving every worker a disjoint task and file scope.
-- Keep workers stateless between tasks: when a worker finishes, integrate or
+- Do not leave workers idle. Either give a worker another coherent follow-on
-  reject the result, then clear that worker context before the next assignment.
+  task while its context is still useful, or close it promptly.
+- Reuse worker context only for closely related follow-on tasks in the same
+  local area; otherwise close the worker and start a fresh one with a new
+  minimal packet.
 - Keep prompts dense with the exact project/task context workers need so they do
   not spend tokens re-reading repo docs or re-parsing broad areas.
 - Keep communication terse: no fillers, no cheerleading, no narrative padding.
@@ -40,7 +46,8 @@ The coordinator is the main agent in the user thread. The coordinator owns:
 - running final validation
 - updating docs/debt/tasks
 - committing and pushing the verified slice
-- resetting worker context after each completed or rejected task
+- either assigning coherent follow-on work to an active worker or closing that
+  worker once its useful context window is over
 
 The coordinator should keep local work minimal. It may do a quick blocking
 check before delegation, such as reading task rows, checking git status, or
@@ -49,13 +56,16 @@ code or test work, delegate it directly to a worker whenever the scope can be
 made clear.
 
 The coordinator must front-load context. Workers should not be told to "read
-the roadmap" or "inspect the repo" unless that is the task. The coordinator is
+the roadmap", "read AGENTS.md", or "inspect the repo" unless that is the task.
-responsible for extracting and passing:
+The coordinator is responsible for extracting and passing:
 
 - task ids and done checks
 - debt ids and removal conditions that matter
 - exact write scope and allowed read scope
 - required validation commands
+- the specific build/test preset or target names the worker may need
+- the exact code-exploration tools to use for the slice, such as `rg` or the
+  compiler-aware `clangd_nav.py` helper
 - relevant file paths, code references, and current behavior notes
 - any repo rules or user constraints that materially affect the task
 
@@ -68,6 +78,11 @@ Workers do not own repo discovery. They start from the coordinator-provided
 context packet and stay inside the assigned scope unless they hit a blocker that
 requires a narrow follow-up question.
 
+Workers may be kept alive for more than one assignment only when the next task
+is coherent with the current one: same subsystem, overlapping read scope,
+similar validation path, and no avoidable context rebuild. Do not keep workers
+alive just because a slot is available.
+
 Every worker and explorer must be told:
 
 - this repository may have other agents working in parallel
@@ -80,13 +95,13 @@ Every worker and explorer must be told:
 
 | Work Type | Model | Reasoning Effort | Use |
 | --- | --- | --- | --- |
-| Coordinator orchestration and integration | inherited main-thread model or `gpt-5.4-mini` | `low` or `medium` | Scope selection, task routing, conflict checks, validation, docs/debt updates, commits, pushes. |
+| Coordinator orchestration and integration | `gpt-5.4` | `low` or `medium` | Scope selection, task routing, conflict checks, validation, docs/debt updates, commits, pushes, and worker packet preparation. |
-| Direct worker coding task | `gpt-5.3-codex-spark` | `medium` | Bounded implementation in known files with coordinator-supplied context. |
+| Direct worker coding task | `gpt-5.4-mini` | `medium` | Bounded implementation in known files with coordinator-supplied context. |
-| Direct worker lookup or inventory | `gpt-5.3-codex-spark` | `low` or `medium` | `rg` inventory, file ownership map, simple grep-based answers. |
+| Direct worker lookup or inventory | `gpt-5.4-mini` | `low` or `medium` | `rg` inventory, file ownership map, simple grep-based answers. |
-| Mechanical docs cleanup | `gpt-5.3-codex-spark` | `low` | Formatting, table updates, command normalization. |
+| Mechanical docs cleanup | `gpt-5.4-mini` | `low` | Formatting, table updates, command normalization. |
 | Coordinator-only escalation | inherited higher model only when explicitly justified | inherited | Resolve architecture ambiguity, conflict integration, or task decomposition failures without adding a captain layer. |
 
-Workers default to `gpt-5.3-codex-spark`. If a task looks too broad or risky
+Workers default to `gpt-5.4-mini`. If a task looks too broad or risky
 for that model, the coordinator should decompose it further or keep the narrow
 integration step locally instead of inserting an extra management tier.
 
@@ -96,33 +111,46 @@ integration step locally instead of inserting an extra management tier.
   ids, validation commands, and only the necessary excerpts.
 - Use `fork_context=true` only when prior conversation details are essential and
   not already captured in the worker prompt.
+- Do not let the coordinator thread drift toward compaction. Once a verified
+  slice is committed and pushed, prefer a fresh thread for the next slice if
+  the remaining context is no longer tight.
 - Do not paste large logs into prompts. Point workers at log paths and ask for
   the smallest relevant excerpt.
-- Do not ask workers to broadly read `AGENTS.md`, the roadmap, or the debt log.
+- Do not ask workers to broadly read `AGENTS.md`, the roadmap, the debt log, or
-  Summarize the exact rules and rows they need.
+  other repo-wide docs. Summarize the exact rules and rows they need.
+- Default worker context to a minimal operating packet: task id, assigned file
+  scope, build command, test command, and code-exploration command hints.
 - Keep worker prompts compact but complete. Shorter is good only if it still
   removes the need for worker-side repo rediscovery.
 - Ask for compact final reports: changed files, result, validation, blockers,
   next recommendation.
-- Close completed workers after their results are integrated or rejected.
+- Keep active workers busy with another coherent task when that is cheaper than
+  restarting context; otherwise close them immediately after integration.
+- Close workers that are done, blocked, or no longer have a strong context
+  advantage so they do not accumulate and saturate worker slots.
 - Prefer the smallest number of concurrent workers that keeps disjoint work
   moving.
 - Use rolling integration: wait for whichever worker finishes first, process the
-  result, then launch the next disjoint worker with a fresh context window.
+  result, then either reuse that worker for the next coherent slice or close it
+  before launching a fresh worker for unrelated work.
+- Prefer committed repo state over chat history as the handoff mechanism between
+  slices so worker and coordinator prompts stay short.
 
 ## Delegation Flow
 
 1. Coordinator picks one or more `Ready` tasks from
    `docs/modernization/tasks.md` with disjoint write scopes.
-2. Coordinator splits each task into direct worker-sized units.
+2. Coordinator splits each task into direct worker-sized units, grouping
+   coherent follow-on work when one worker can finish it efficiently without
+   broadening scope.
 3. Coordinator prepares a context packet for each worker with the exact task
    requirements, file scope, validation commands, and relevant project details.
-4. Coordinator assigns the task directly to a `gpt-5.3-codex-spark` worker or
+4. Coordinator assigns the task directly to a `gpt-5.4-mini` worker or
    explorer.
 5. Worker returns changed files, validation, blockers, and any narrow
    integration notes.
-6. Coordinator reviews for scope conflicts, integrates the result, and clears
+6. Coordinator reviews for scope conflicts, integrates the result, and decides
-   that worker context before assigning the next task.
+   whether to give that same worker another coherent task or close it.
 7. Coordinator runs the listed validation command or the quiet checkpoint
    wrapper for each integrated slice.
 8. Coordinator updates `tasks.md`, `debt.md`, and `roadmap.md` if task state or
@@ -132,7 +160,7 @@ integration step locally instead of inserting an extra management tier.
 ## Coordinator Prompt Template For A Worker
 
 ```text
-You are a `gpt-5.3-codex-spark` worker on PanoPainter. Other agents may be
+You are a `gpt-5.4-mini` worker on PanoPainter. Other agents may be
 editing nearby files; do not revert unrelated changes.
 
 Task source: docs/modernization/tasks.md task(s) <TASK-ID-LIST>.
@@ -142,15 +170,24 @@ Debt ids: <DEBT-LIST>.
 Write scope: <FILES/DIRS ONLY>.
 Read scope: <FILES/DIRS>.
 Validation: <COMMANDS>.
+Code exploration: <RG OR CLANGD_NAV COMMANDS TO USE>.
 
 Repo constraints you must follow:
 - <ONLY THE RELEVANT RULES>
 
-Current context you should rely on instead of broad repo/doc review:
+Minimal context you should rely on instead of broad repo/doc review:
 - <CURRENT BEHAVIOR NOTE>
 - <RELEVANT FILE OR SYMBOL NOTE>
 - <WHY THIS SLICE IS SAFE / WHAT MUST NOT CHANGE>
 
+Do not read repo-wide docs unless this task explicitly requires it. Use only
+the supplied context, the listed file scope, and the build/test/code-exploration
+commands above.
+
+If the coordinator gives you a second task, accept it only when it is coherent
+with the current scope and does not require broad repo rediscovery. Otherwise
+say that a fresh worker should be used.
+
 Make the smallest behavior-preserving change that satisfies the done checks.
 Do not spend tokens on broad document review or inventory outside the assigned
 scope unless the task explicitly requires it. If the task is larger than
@@ -187,6 +224,11 @@ Return only:
 - Focused validation for the task passed or the failure is documented.
 - `docs/modernization/debt.md` changed when debt was narrowed or closed.
 - `docs/modernization/tasks.md` score changed only for `Done` tasks.
-- Each worker result was integrated before that worker context was reused.
+- Each worker result was integrated before that worker was reused.
+- Reused workers only handled coherent follow-on tasks with a real context
+  advantage.
+- Done or blocked workers were closed instead of being left idle.
+- The coordinator did not carry unnecessary stale history when a fresh thread
+  would have been cheaper than compaction.
 - The commit contains one coherent slice.
 - The branch was pushed.

src/app.cpp

@@ -32,10 +32,6 @@
 
 App* App::I = nullptr; // singleton
 
-std::deque<AppTask> App::render_tasklist;
-std::mutex App::render_task_mutex;
-std::condition_variable App::render_cv;
-
 namespace {
 
 pp::app::CanvasToolMode canvas_tool_mode_from_canvas_mode(kCanvasMode mode) noexcept
@@ -130,17 +126,6 @@ void apply_app_scissor_test(bool enabled)
 }
 
 }
-std::thread App::render_thread;
-std::thread::id App::render_thread_id;
-bool App::render_running = false;
-
-std::deque<AppTask> App::ui_tasklist;
-std::mutex App::ui_task_mutex;
-std::condition_variable App::ui_cv;
-std::thread App::ui_thread;
-std::thread::id App::ui_thread_id;
-bool App::ui_running = false;
-
 void App::create()
 {
     const auto initial_surface = pp::app::plan_app_initial_surface();
@@ -353,7 +338,7 @@ void App::async_redraw()
     if (plan.set_redraw)
         redraw = true;
     if (plan.notify_ui)
-        ui_cv.notify_all();
+        runtime_.notify_ui_worker();
 }
 
 void App::async_end()
@@ -699,252 +684,40 @@ void App::rec_loop()
 
 void App::render_thread_tick()
 {
-    static uint32_t count = 0;
+    runtime_.render_thread_tick(*this);
-    render_thread_id = std::this_thread::get_id();
-    std::deque<AppTask> working_list;
-    pp::app::AppQueueDrainPlan drain_plan;
-
-    // move the task list locally to free the queue for other threads
-    {
-        std::unique_lock<std::mutex> lock(render_task_mutex);
-        drain_plan = pp::app::plan_app_render_queue_drain(render_tasklist.size());
-        render_running = drain_plan.mark_running;
-        if (!drain_plan.drain_tasks)
-            return;
-        working_list = std::move(render_tasklist);
-    }
-
-    // execute the tasks
-    if (drain_plan.wrap_in_render_context)
-    {
-        async_start();
-        while (!working_list.empty())
-        {
-            //LOG("render task %d", count);
-            //LOG("task %s", working_list.front().name.c_str());
-            count++;
-            working_list.front()();
-            working_list.pop_front();
-        }
-        async_end();
-    }
 }
 
 void App::render_thread_main()
 {
-    BT_SetTerminate();
+    runtime_.render_thread_main(*this);
-
-    uint32_t count = 0;
-    render_thread_id = std::this_thread::get_id();
-    render_running = pp::app::plan_app_thread_start().mark_running;
-    while (render_running)
-    {
-        std::deque<AppTask> working_list;
-        pp::app::AppQueueDrainPlan drain_plan;
-
-        // move the task list locally to free the queue for other threads
-        {
-            std::unique_lock<std::mutex> lock(render_task_mutex);
-            render_cv.wait(lock, [this] { return render_tasklist.empty() && render_running ? false : true; });
-            drain_plan = pp::app::plan_app_render_queue_drain(render_tasklist.size());
-            working_list = std::move(render_tasklist);
-        }
-
-        // execute the tasks
-        if (drain_plan.wrap_in_render_context)
-        {
-            async_start();
-            while (!working_list.empty())
-            {
-                //LOG("render task %d", count);
-                //LOG("task %s", working_list.front().name.c_str());
-                count++;
-                working_list.front()();
-                working_list.pop_front();
-            }
-            async_end();
-        }
-    }
 }
 
 void App::ui_thread_tick()
 {
-    ui_thread_id = std::this_thread::get_id();
+    runtime_.ui_thread_tick(*this);
-
-    std::deque<AppTask> working_list;
-    pp::app::AppUiTickPlan tick_plan;
-
-    // move the task list locally to free the queue for other threads
-    {
-        std::unique_lock<std::mutex> lock(ui_task_mutex);
-        tick_plan = pp::app::plan_app_ui_thread_tick(ui_tasklist.size(), redraw);
-        ui_running = tick_plan.mark_running;
-        working_list = std::move(ui_tasklist);
-    }
-
-    // execute the tasks
-    if (tick_plan.execute_tasks)
-    {
-        while (!working_list.empty())
-        {
-            //LOG("ui task %d", count);
-            working_list.front()();
-            working_list.pop_front();
-        }
-    }
-
-    if (tick_plan.tick_app)
-        tick(0);
-
-    const auto redraw_plan = pp::app::plan_app_ui_loop_redraw(redraw, 0);
-    if (redraw_plan.enqueue_render_frame)
-    {
-        if (redraw_plan.update_before_render)
-            update(0);
-        render_task([this]
-        {
-            bind_default_render_target();
-            clear();
-            draw(0);
-            async_swap();
-        });
-    }
 }
 
 void App::ui_thread_main()
 {
-    BT_SetTerminate();
+    runtime_.ui_thread_main(*this);
-
-    uint32_t count = 0;
-    ui_thread_id = std::this_thread::get_id();
-    ui_running = pp::app::plan_app_thread_start().mark_running;
-
-    attach_ui_thread();
-
-    LOG("ui thread init()");
-    init();
-
-    auto t_start = std::chrono::high_resolution_clock::now();
-    float t_frame = 0;
-    float t_fps_counter = 0;
-    float t_reloader = 0;
-    int rendered_frames = 0;
-    while (ui_running)
-    {
-        std::deque<AppTask> working_list;
-
-        // move the task list locally to free the queue for other threads
-        {
-            std::unique_lock<std::mutex> lock(ui_task_mutex);
-            ui_cv.wait_for(lock, std::chrono::milliseconds(idle_ms), 
-                [this] { return ui_tasklist.empty() && ui_running ? false : true; });
-            working_list = std::move(ui_tasklist);
-        }
-
-        // execute the tasks
-        if (!working_list.empty())
-        {
-            while (!working_list.empty())
-            {
-                //LOG("ui task %d", count);
-                count++;
-                working_list.front()();
-                working_list.pop_front();
-            }
-        }
-
-        auto t_now = std::chrono::high_resolution_clock::now();
-        float dt = std::chrono::duration<float>(t_now - t_start).count();
-        t_start = t_now;
-
-        const auto timer_plan = pp::app::plan_app_ui_loop_timers(
-            dt,
-            t_frame,
-            t_fps_counter,
-            t_reloader,
-            rendered_frames,
-            platform_enables_live_asset_reloading());
-        if (timer_plan) {
-            if (timer_plan.value().update_platform_frame)
-                update_platform_frame(dt);
-            t_frame = timer_plan.value().frame_accumulator;
-            t_fps_counter = timer_plan.value().fps_accumulator;
-            t_reloader = timer_plan.value().reload_accumulator;
-            rendered_frames = timer_plan.value().rendered_frames_after_report;
-
-            if (timer_plan.value().report_rendered_frames)
-                report_rendered_frames(timer_plan.value().reported_frame_count);
-
-            if (timer_plan.value().check_live_asset_reload) {
-                if (ShaderManager::reload())
-                {
-                    stroke->update_controls();
-                    redraw = true;
-                }
-                if (layout.reload())
-                    redraw = true;
-                if (layout_designer.reload())
-                    redraw = true;
-            }
-        }
-
-        const auto redraw_plan = pp::app::plan_app_ui_loop_redraw(redraw, rendered_frames);
-        if (redraw_plan.tick_app)
-            tick(dt);
-
-        if (redraw_plan.enqueue_render_frame)
-        {
-            if (redraw_plan.update_before_render)
-                update(t_frame);
-            render_task([this, t_frame]
-            {
-                bind_default_render_target();
-                clear();
-                draw(t_frame);
-                async_swap();
-            });
-            if (redraw_plan.reset_frame_accumulator)
-                t_frame = 0;
-            rendered_frames = redraw_plan.rendered_frames;
-        }
-    }
-    detach_ui_thread();
 }
 
 void App::render_thread_start()
 {
-    const auto plan = pp::app::plan_app_thread_start();
+    runtime_.render_thread_start(*this);
-    if (plan.start_thread)
-        render_thread = std::thread(&App::render_thread_main, this);
-    render_running = plan.mark_running;
 }
 
 void App::render_thread_stop()
 {
-    const auto plan = pp::app::plan_app_thread_stop(render_thread.joinable());
+    runtime_.render_thread_stop();
-    if (plan.mark_not_running)
-        render_running = false;
-    if (plan.notify_worker)
-        render_cv.notify_all();
-    if (plan.join_thread)
-        render_thread.join();
 }
 
 void App::ui_thread_start()
 {
-    const auto plan = pp::app::plan_app_thread_start();
+    runtime_.ui_thread_start(*this);
-    if (plan.start_thread)
-        ui_thread = std::thread(&App::ui_thread_main, this);
-    ui_running = plan.mark_running;
 }
 
 void App::ui_thread_stop()
 {
-    const auto plan = pp::app::plan_app_thread_stop(ui_thread.joinable());
+    runtime_.ui_thread_stop();
-    if (plan.mark_not_running)
-        ui_running = false;
-    if (plan.notify_worker)
-        ui_cv.notify_all();
-    if (plan.join_thread)
-        ui_thread.join();
 }

src/app.h

@@ -25,6 +25,7 @@
 #include "layout.h"
 #include "app_core/document_session.h"
 #include "app_core/app_thread.h"
+#include "app_runtime.h"
 
 namespace pp::platform {
 class PlatformServices;
@@ -76,21 +77,6 @@ struct VRController
     virtual float get_trigger_value() const { return 1.f; }
 };
 
-struct AppTask : public std::packaged_task<void()>
-{
-    size_t task_id;
-#ifdef _DEBUG
-    std::string name;
-#endif
-    template<typename F> AppTask(F f) : std::packaged_task<void()>(f)
-    {
-        task_id = typeid(f).hash_code();
-#ifdef _DEBUG
-        name = typeid(f).name();
-#endif
-    }
-};
-
 class App
 {
 public:
@@ -345,16 +331,13 @@ public:
 
     void cmd_convert(std::string pano_path, std::string out_path);
 
+    AppRuntime& runtime() noexcept { return runtime_; }
+    const AppRuntime& runtime() const noexcept { return runtime_; }
+
     //////////////////////////////////////////////////////////////////////////
     // RENDER THREAD
     //////////////////////////////////////////////////////////////////////////
 
-    static std::deque<AppTask> render_tasklist;
-    static std::mutex render_task_mutex;
-    static std::condition_variable render_cv;
-    static std::thread render_thread;
-    static std::thread::id render_thread_id;
-    static bool render_running;
     void render_thread_tick();
     void render_thread_main();
     void render_thread_start();
@@ -362,94 +345,30 @@ public:
 
     bool is_render_thread()
     {
-        return std::this_thread::get_id() == render_thread_id;
+        return runtime_.is_render_thread();
     }
 
-    // don't capture a reference to this ptr as the object may be destroyed
-    // by the time the task is executed
     template<typename T>
     std::future<void> render_task_async(T task, bool unique = false)
     {
-        AppTask pt(task);
+        return runtime_.render_task_async(std::move(task), unique);
-        auto f = pt.get_future();
-        const auto dispatch = pp::app::plan_app_task_dispatch(
-            is_render_thread(),
-            unique,
-            0U,
-            render_running,
-            false,
-            false);
-        if (dispatch.execute_immediately)
-        {
-            pt();
-        }
-        else if (dispatch.queue_task)
-        {
-            {
-                std::lock_guard<std::mutex> lock(render_task_mutex);
-                const auto queue_dispatch = pp::app::plan_app_task_dispatch(
-                    false,
-                    unique,
-                    render_tasklist.size(),
-                    render_running,
-                    false,
-                    false);
-                // remove any previously queued task from the same lambda
-                if (queue_dispatch.remove_matching_unique_task)
-                    render_tasklist.erase(std::remove_if(render_tasklist.begin(), render_tasklist.end(),
-                        [id = pt.task_id](AppTask const& t){ return t.task_id == id; }), render_tasklist.end());
-                render_tasklist.push_back(std::move(pt));
-            }
-            if (dispatch.notify_worker)
-                render_cv.notify_all();
-        }
-        return f;
     }
 
     template<typename T>
     void render_task(T task)
     {
-        AppTask pt(task);
+        runtime_.render_task(std::move(task));
-        auto f = pt.get_future();
-        const auto dispatch = pp::app::plan_app_task_dispatch(
-            is_render_thread(),
-            false,
-            0U,
-            render_running,
-            true,
-            false);
-        if (dispatch.execute_immediately)
-        {
-            pt();
-        }
-        else if (dispatch.queue_task)
-        {
-            {
-                std::lock_guard<std::mutex> lock(render_task_mutex);
-                render_tasklist.push_back(std::move(pt));
-            }
-            if (dispatch.notify_worker)
-                render_cv.notify_all();
-        }
-        if (dispatch.wait_for_completion)
-            f.get();
     }
 
     void render_sync()
     {
-        render_task([] {});
+        runtime_.render_sync();
     }
 
     //////////////////////////////////////////////////////////////////////////
     // UI THREAD
     //////////////////////////////////////////////////////////////////////////
 
-    static std::deque<AppTask> ui_tasklist;
-    static std::mutex ui_task_mutex;
-    static std::condition_variable ui_cv;
-    static std::thread ui_thread;
-    static std::thread::id ui_thread_id;
-    static bool ui_running;
     void ui_thread_tick();
     void ui_thread_main();
     void ui_thread_start();
@@ -457,83 +376,29 @@ public:
 
     bool is_ui_thread()
     {
-        return std::this_thread::get_id() == ui_thread_id;
+        return runtime_.is_ui_thread();
     }
 
-    // don't capture a reference to this ptr as the object may be destroyed
-    // by the time the task is executed
     template<typename T>
     std::future<void> ui_task_async(T task, bool unique = false)
     {
-        AppTask pt(task);
+        return runtime_.ui_task_async(std::move(task), unique);
-        auto f = pt.get_future();
-        const auto dispatch = pp::app::plan_app_task_dispatch(
-            is_ui_thread(),
-            unique,
-            0U,
-            ui_running,
-            false,
-            false);
-        if (dispatch.execute_immediately)
-        {
-            pt();
-        }
-        else if (dispatch.queue_task)
-        {
-            {
-                std::lock_guard<std::mutex> lock(ui_task_mutex);
-                const auto queue_dispatch = pp::app::plan_app_task_dispatch(
-                    false,
-                    unique,
-                    ui_tasklist.size(),
-                    ui_running,
-                    false,
-                    false);
-                // remove any previously queued task from the same lambda
-                if (queue_dispatch.remove_matching_unique_task)
-                    ui_tasklist.erase(std::remove_if(ui_tasklist.begin(), ui_tasklist.end(), 
-                        [id = pt.task_id](AppTask const& t){ return t.task_id == id; }), ui_tasklist.end());
-                ui_tasklist.push_back(std::move(pt));
-            }
-            if (dispatch.notify_worker)
-                ui_cv.notify_all();
-        }
-        return f;
     }
 
     template<typename T>
     void ui_task(T task)
     {
-        AppTask pt(task);
+        runtime_.ui_task(std::move(task));
-        auto f = pt.get_future();
+        if (runtime_.request_redraw())
-        const auto dispatch = pp::app::plan_app_task_dispatch(
-            is_ui_thread(),
-            false,
-            0U,
-            ui_running,
-            true,
-            true);
-        if (dispatch.execute_immediately)
-        {
-            pt();
-        }
-        else if (dispatch.queue_task)
-        {
-            {
-                std::lock_guard<std::mutex> lock(ui_task_mutex);
-                ui_tasklist.push_back(std::move(pt));
-            }
-            if (dispatch.notify_worker)
-                ui_cv.notify_all();
-        }
-        if (dispatch.wait_for_completion)
-            f.get();
-        if (dispatch.request_redraw)
             redraw = true;
+        runtime_.clear_request_redraw();
     }
 
     void ui_sync()
     {
-        ui_task([] {});
+        runtime_.ui_sync();
     }
+
+private:
+    AppRuntime runtime_;
 };

src/app_cloud.cpp

@@ -19,7 +19,7 @@ void App::cloud_upload()
 
 void App::cloud_upload_all()
 {
-    std::thread([this] {
+    pp::panopainter::queue_legacy_cloud_worker_task([this] {
         BT_SetTerminate();
 
         auto names = Asset::list_files(data_path, ".*\\.ppi");
@@ -28,7 +28,7 @@ void App::cloud_upload_all()
         const auto status = pp::panopainter::execute_legacy_cloud_bulk_upload_plan(*this, plan);
         if (!status.ok())
             LOG("Cloud bulk upload action failed: %s", status.message);
-    }).detach();
+    });
 }
 
 void App::cloud_browse()

src/app_runtime.cpp

@@ -0,0 +1,258 @@
+#include "pch.h"
+#include "app_runtime.h"
+
+#include "app.h"
+
+bool AppRuntime::is_render_thread() const noexcept
+{
+    return std::this_thread::get_id() == render_thread_id_;
+}
+
+bool AppRuntime::is_ui_thread() const noexcept
+{
+    return std::this_thread::get_id() == ui_thread_id_;
+}
+
+void AppRuntime::notify_render_worker() noexcept
+{
+    render_cv_.notify_all();
+}
+
+void AppRuntime::notify_ui_worker() noexcept
+{
+    ui_cv_.notify_all();
+}
+
+void AppRuntime::render_thread_tick(App& app)
+{
+    static uint32_t count = 0;
+    render_thread_id_ = std::this_thread::get_id();
+    std::deque<AppTask> working_list;
+    pp::app::AppQueueDrainPlan drain_plan;
+
+    {
+        std::unique_lock<std::mutex> lock(render_task_mutex_);
+        drain_plan = pp::app::plan_app_render_queue_drain(render_tasklist_.size());
+        render_running_ = drain_plan.mark_running;
+        if (!drain_plan.drain_tasks)
+            return;
+        working_list = std::move(render_tasklist_);
+    }
+
+    if (drain_plan.wrap_in_render_context)
+    {
+        app.async_start();
+        while (!working_list.empty())
+        {
+            count++;
+            working_list.front()();
+            working_list.pop_front();
+        }
+        app.async_end();
+    }
+}
+
+void AppRuntime::render_thread_main(App& app)
+{
+    BT_SetTerminate();
+
+    render_thread_id_ = std::this_thread::get_id();
+    render_running_ = pp::app::plan_app_thread_start().mark_running;
+    while (render_running_)
+    {
+        std::deque<AppTask> working_list;
+        pp::app::AppQueueDrainPlan drain_plan;
+
+        {
+            std::unique_lock<std::mutex> lock(render_task_mutex_);
+            render_cv_.wait(lock, [this] { return render_tasklist_.empty() && render_running_ ? false : true; });
+            drain_plan = pp::app::plan_app_render_queue_drain(render_tasklist_.size());
+            working_list = std::move(render_tasklist_);
+        }
+
+        if (drain_plan.wrap_in_render_context)
+        {
+            app.async_start();
+            while (!working_list.empty())
+            {
+                working_list.front()();
+                working_list.pop_front();
+            }
+            app.async_end();
+        }
+    }
+}
+
+void AppRuntime::ui_thread_tick(App& app)
+{
+    ui_thread_id_ = std::this_thread::get_id();
+
+    std::deque<AppTask> working_list;
+    pp::app::AppUiTickPlan tick_plan;
+
+    {
+        std::unique_lock<std::mutex> lock(ui_task_mutex_);
+        tick_plan = pp::app::plan_app_ui_thread_tick(ui_tasklist_.size(), app.redraw);
+        ui_running_ = tick_plan.mark_running;
+        working_list = std::move(ui_tasklist_);
+    }
+
+    if (tick_plan.execute_tasks)
+    {
+        while (!working_list.empty())
+        {
+            working_list.front()();
+            working_list.pop_front();
+        }
+    }
+
+    if (tick_plan.tick_app)
+        app.tick(0);
+
+    const auto redraw_plan = pp::app::plan_app_ui_loop_redraw(app.redraw, 0);
+    if (redraw_plan.enqueue_render_frame)
+    {
+        if (redraw_plan.update_before_render)
+            app.update(0);
+        app.render_task([&app]
+        {
+            app.bind_default_render_target();
+            app.clear();
+            app.draw(0);
+            app.async_swap();
+        });
+    }
+}
+
+void AppRuntime::ui_thread_main(App& app)
+{
+    BT_SetTerminate();
+
+    ui_thread_id_ = std::this_thread::get_id();
+    ui_running_ = pp::app::plan_app_thread_start().mark_running;
+
+    app.attach_ui_thread();
+
+    LOG("ui thread init()");
+    app.init();
+
+    auto t_start = std::chrono::high_resolution_clock::now();
+    float t_frame = 0;
+    float t_fps_counter = 0;
+    float t_reloader = 0;
+    int rendered_frames = 0;
+    while (ui_running_)
+    {
+        std::deque<AppTask> working_list;
+
+        {
+            std::unique_lock<std::mutex> lock(ui_task_mutex_);
+            ui_cv_.wait_for(lock, std::chrono::milliseconds(app.idle_ms), 
+                [this] { return ui_tasklist_.empty() && ui_running_ ? false : true; });
+            working_list = std::move(ui_tasklist_);
+        }
+
+        if (!working_list.empty())
+        {
+            while (!working_list.empty())
+            {
+                working_list.front()();
+                working_list.pop_front();
+            }
+        }
+
+        auto t_now = std::chrono::high_resolution_clock::now();
+        float dt = std::chrono::duration<float>(t_now - t_start).count();
+        t_start = t_now;
+
+        const auto timer_plan = pp::app::plan_app_ui_loop_timers(
+            dt,
+            t_frame,
+            t_fps_counter,
+            t_reloader,
+            rendered_frames,
+            app.platform_enables_live_asset_reloading());
+        if (timer_plan) {
+            if (timer_plan.value().update_platform_frame)
+                app.update_platform_frame(dt);
+            t_frame = timer_plan.value().frame_accumulator;
+            t_fps_counter = timer_plan.value().fps_accumulator;
+            t_reloader = timer_plan.value().reload_accumulator;
+            rendered_frames = timer_plan.value().rendered_frames_after_report;
+
+            if (timer_plan.value().report_rendered_frames)
+                app.report_rendered_frames(timer_plan.value().reported_frame_count);
+
+            if (timer_plan.value().check_live_asset_reload) {
+                if (ShaderManager::reload())
+                {
+                    app.stroke->update_controls();
+                    app.redraw = true;
+                }
+                if (app.layout.reload())
+                    app.redraw = true;
+                if (app.layout_designer.reload())
+                    app.redraw = true;
+            }
+        }
+
+        const auto redraw_plan = pp::app::plan_app_ui_loop_redraw(app.redraw, rendered_frames);
+        if (redraw_plan.tick_app)
+            app.tick(dt);
+
+        if (redraw_plan.enqueue_render_frame)
+        {
+            if (redraw_plan.update_before_render)
+                app.update(t_frame);
+            app.render_task([&app, t_frame]
+            {
+                app.bind_default_render_target();
+                app.clear();
+                app.draw(t_frame);
+                app.async_swap();
+            });
+            if (redraw_plan.reset_frame_accumulator)
+                t_frame = 0;
+            rendered_frames = redraw_plan.rendered_frames;
+        }
+    }
+    app.detach_ui_thread();
+}
+
+void AppRuntime::render_thread_start(App& app)
+{
+    const auto plan = pp::app::plan_app_thread_start();
+    if (plan.start_thread)
+        render_thread_ = std::thread(&AppRuntime::render_thread_main, this, std::ref(app));
+    render_running_ = plan.mark_running;
+}
+
+void AppRuntime::render_thread_stop()
+{
+    const auto plan = pp::app::plan_app_thread_stop(render_thread_.joinable());
+    if (plan.mark_not_running)
+        render_running_ = false;
+    if (plan.notify_worker)
+        render_cv_.notify_all();
+    if (plan.join_thread)
+        render_thread_.join();
+}
+
+void AppRuntime::ui_thread_start(App& app)
+{
+    const auto plan = pp::app::plan_app_thread_start();
+    if (plan.start_thread)
+        ui_thread_ = std::thread(&AppRuntime::ui_thread_main, this, std::ref(app));
+    ui_running_ = plan.mark_running;
+}
+
+void AppRuntime::ui_thread_stop()
+{
+    const auto plan = pp::app::plan_app_thread_stop(ui_thread_.joinable());
+    if (plan.mark_not_running)
+        ui_running_ = false;
+    if (plan.notify_worker)
+        ui_cv_.notify_all();
+    if (plan.join_thread)
+        ui_thread_.join();
+}

src/app_runtime.h

@@ -0,0 +1,216 @@
+#pragma once
+
+#include "app_core/app_thread.h"
+
+#include <algorithm>
+#include <condition_variable>
+#include <cstddef>
+#include <deque>
+#include <functional>
+#include <future>
+#include <mutex>
+#include <string>
+#include <typeinfo>
+#include <utility>
+#include <thread>
+
+class App;
+
+struct AppTask : public std::packaged_task<void()>
+{
+    size_t task_id;
+#ifdef _DEBUG
+    std::string name;
+#endif
+    template<typename F> AppTask(F f) : std::packaged_task<void()>(f)
+    {
+        task_id = typeid(f).hash_code();
+#ifdef _DEBUG
+        name = typeid(f).name();
+#endif
+    }
+};
+
+class AppRuntime
+{
+public:
+    [[nodiscard]] bool is_render_thread() const noexcept;
+    [[nodiscard]] bool is_ui_thread() const noexcept;
+    [[nodiscard]] bool request_redraw() const noexcept { return request_redraw_; }
+    void clear_request_redraw() noexcept { request_redraw_ = false; }
+
+    void notify_render_worker() noexcept;
+    void notify_ui_worker() noexcept;
+
+    void render_thread_tick(App& app);
+    void render_thread_main(App& app);
+    void render_thread_start(App& app);
+    void render_thread_stop();
+
+    void ui_thread_tick(App& app);
+    void ui_thread_main(App& app);
+    void ui_thread_start(App& app);
+    void ui_thread_stop();
+
+    template<typename T>
+    std::future<void> render_task_async(T task, bool unique = false)
+    {
+        AppTask pt(task);
+        auto f = pt.get_future();
+        const auto dispatch = pp::app::plan_app_task_dispatch(
+            is_render_thread(),
+            unique,
+            0U,
+            render_running_,
+            false,
+            false);
+        if (dispatch.execute_immediately)
+        {
+            pt();
+        }
+        else if (dispatch.queue_task)
+        {
+            {
+                std::lock_guard<std::mutex> lock(render_task_mutex_);
+                const auto queue_dispatch = pp::app::plan_app_task_dispatch(
+                    false,
+                    unique,
+                    render_tasklist_.size(),
+                    render_running_,
+                    false,
+                    false);
+                if (queue_dispatch.remove_matching_unique_task)
+                    render_tasklist_.erase(std::remove_if(render_tasklist_.begin(), render_tasklist_.end(),
+                        [id = pt.task_id](AppTask const& t){ return t.task_id == id; }), render_tasklist_.end());
+                render_tasklist_.push_back(std::move(pt));
+            }
+            if (dispatch.notify_worker)
+                render_cv_.notify_all();
+        }
+        return f;
+    }
+
+    template<typename T>
+    void render_task(T task)
+    {
+        AppTask pt(task);
+        auto f = pt.get_future();
+        const auto dispatch = pp::app::plan_app_task_dispatch(
+            is_render_thread(),
+            false,
+            0U,
+            render_running_,
+            true,
+            false);
+        if (dispatch.execute_immediately)
+        {
+            pt();
+        }
+        else if (dispatch.queue_task)
+        {
+            {
+                std::lock_guard<std::mutex> lock(render_task_mutex_);
+                render_tasklist_.push_back(std::move(pt));
+            }
+            if (dispatch.notify_worker)
+                render_cv_.notify_all();
+        }
+        if (dispatch.wait_for_completion)
+            f.get();
+    }
+
+    void render_sync()
+    {
+        render_task([] {});
+    }
+
+    template<typename T>
+    std::future<void> ui_task_async(T task, bool unique = false)
+    {
+        AppTask pt(task);
+        auto f = pt.get_future();
+        const auto dispatch = pp::app::plan_app_task_dispatch(
+            is_ui_thread(),
+            unique,
+            0U,
+            ui_running_,
+            false,
+            false);
+        if (dispatch.execute_immediately)
+        {
+            pt();
+        }
+        else if (dispatch.queue_task)
+        {
+            {
+                std::lock_guard<std::mutex> lock(ui_task_mutex_);
+                const auto queue_dispatch = pp::app::plan_app_task_dispatch(
+                    false,
+                    unique,
+                    ui_tasklist_.size(),
+                    ui_running_,
+                    false,
+                    false);
+                if (queue_dispatch.remove_matching_unique_task)
+                    ui_tasklist_.erase(std::remove_if(ui_tasklist_.begin(), ui_tasklist_.end(), 
+                        [id = pt.task_id](AppTask const& t){ return t.task_id == id; }), ui_tasklist_.end());
+                ui_tasklist_.push_back(std::move(pt));
+            }
+            if (dispatch.notify_worker)
+                ui_cv_.notify_all();
+        }
+        return f;
+    }
+
+    template<typename T>
+    void ui_task(T task)
+    {
+        AppTask pt(task);
+        auto f = pt.get_future();
+        const auto dispatch = pp::app::plan_app_task_dispatch(
+            is_ui_thread(),
+            false,
+            0U,
+            ui_running_,
+            true,
+            true);
+        if (dispatch.execute_immediately)
+        {
+            pt();
+        }
+        else if (dispatch.queue_task)
+        {
+            {
+                std::lock_guard<std::mutex> lock(ui_task_mutex_);
+                ui_tasklist_.push_back(std::move(pt));
+            }
+            if (dispatch.notify_worker)
+                ui_cv_.notify_all();
+        }
+        if (dispatch.wait_for_completion)
+            f.get();
+        if (dispatch.request_redraw)
+            request_redraw_ = true;
+    }
+
+    void ui_sync()
+    {
+        ui_task([] {});
+    }
+
+private:
+    std::deque<AppTask> render_tasklist_;
+    std::mutex render_task_mutex_;
+    std::condition_variable render_cv_;
+    std::thread render_thread_;
+    std::thread::id render_thread_id_;
+    bool render_running_ = false;
+
+    std::deque<AppTask> ui_tasklist_;
+    std::mutex ui_task_mutex_;
+    std::condition_variable ui_cv_;
+    std::thread ui_thread_;
+    std::thread::id ui_thread_id_;
+    bool ui_running_ = false;
+    bool request_redraw_ = false;
+};

src/legacy_cloud_services.cpp

@@ -21,6 +21,79 @@ pp::foundation::Status execute_legacy_cloud_download_selection_action(
 
 namespace {
 
+class LegacyCloudWorker final {
+public:
+    LegacyCloudWorker()
+        : worker_([this](std::stop_token stop_token) {
+            run(stop_token);
+        })
+    {
+    }
+
+    ~LegacyCloudWorker()
+    {
+        shutdown();
+    }
+
+    void post(std::function<void()> task)
+    {
+        {
+            std::lock_guard<std::mutex> lock(mutex_);
+            if (stopping_)
+                return;
+            tasks_.push_back(std::move(task));
+        }
+        cv_.notify_one();
+    }
+
+private:
+    void shutdown()
+    {
+        {
+            std::lock_guard<std::mutex> lock(mutex_);
+            stopping_ = true;
+        }
+        cv_.notify_all();
+    }
+
+    void run(std::stop_token stop_token)
+    {
+        for (;;) {
+            std::function<void()> task;
+            {
+                std::unique_lock<std::mutex> lock(mutex_);
+                cv_.wait(lock, [&] {
+                    return stopping_ || stop_token.stop_requested() || !tasks_.empty();
+                });
+                if ((stopping_ || stop_token.stop_requested()) && tasks_.empty())
+                    break;
+                task = std::move(tasks_.front());
+                tasks_.pop_front();
+            }
+
+            if (task) {
+                try {
+                    task();
+                } catch (...) {
+                    LOG("cloud worker task failed");
+                }
+            }
+        }
+    }
+
+    std::mutex mutex_;
+    std::condition_variable cv_;
+    std::deque<std::function<void()>> tasks_;
+    bool stopping_ = false;
+    std::jthread worker_;
+};
+
+LegacyCloudWorker& cloud_worker()
+{
+    static LegacyCloudWorker worker;
+    return worker;
+}
+
 #if WITH_CURL
 int progress_callback_download(void* clientp, curl_off_t dltotal,
     curl_off_t dlnow, curl_off_t ultotal, curl_off_t ulnow)
@@ -233,28 +306,12 @@ void execute_cloud_download_thread(
     execute_cloud_download_flow(app, request);
 }
 
-void launch_cloud_download_thread(
-    App& app,
-    const pp::app::CloudDownloadRequest& request)
-{
-    std::thread([app = &app, request] {
-        execute_cloud_download_thread(*app, request);
-    }).detach();
-}
-
 void execute_cloud_publish_worker(App& app, bool save_before_upload)
 {
     BT_SetTerminate();
     execute_cloud_publish_transfer_and_success_prompt(app, save_before_upload);
 }
 
-void launch_cloud_publish_thread(App& app, bool save_before_upload)
-{
-    std::thread([app = &app, save_before_upload] {
-        execute_cloud_publish_worker(*app, save_before_upload);
-    }).detach();
-}
-
 void wire_cloud_publish_prompt_buttons(
     const std::shared_ptr<NodeMessageBox>& dialog,
     std::function<void()> upload_thread)
@@ -273,7 +330,9 @@ void setup_cloud_publish_prompt(App& app, bool save_before_upload)
     const auto prompt_plan = pp::app::plan_cloud_publish_prompt();
     auto dialog = app.message_box(prompt_plan.title, prompt_plan.message, prompt_plan.show_cancel);
     wire_cloud_publish_prompt_buttons(dialog, [&app, save_before_upload] {
-        launch_cloud_publish_thread(app, save_before_upload);
+        queue_legacy_cloud_worker_task([app = &app, save_before_upload] {
+            execute_cloud_publish_worker(*app, save_before_upload);
+        });
     });
 }
 
@@ -379,7 +438,9 @@ public:
 
     void start_download(const pp::app::CloudDownloadRequest& request) override
     {
-        launch_cloud_download_thread(app_, request);
+        queue_legacy_cloud_worker_task([app = &app_, request] {
+            execute_cloud_download_thread(*app, request);
+        });
     }
 
 private:
@@ -395,6 +456,11 @@ void show_cloud_save_required_warning(App& app)
 
 } // namespace
 
+void queue_legacy_cloud_worker_task(std::function<void()> task)
+{
+    cloud_worker().post(std::move(task));
+}
+
 pp::foundation::Status execute_legacy_cloud_upload_plan(
     App& app,
     const pp::app::CloudUploadPlan& plan)

src/legacy_cloud_services.h

@@ -3,6 +3,8 @@
 #include "app_core/document_cloud.h"
 #include "foundation/result.h"
 
+#include <functional>
+
 class App;
 class NodeDialogCloud;
 
@@ -22,4 +24,6 @@ namespace pp::panopainter {
     pp::app::CloudDownloadSelectionAction action,
     NodeDialogCloud& dialog);
 
+void queue_legacy_cloud_worker_task(std::function<void()> task);
+
 } // namespace pp::panopainter

src/legacy_ui_overlay_services.cpp

@@ -174,6 +174,25 @@ void close_legacy_overlay_handle_ignoring_status(
     (void)close_legacy_overlay_node(anchor, overlay);
 }
 
+void close_legacy_overlay_handle_and_reset(
+    Node& anchor,
+    pp::ui::NodeHandle& overlay) noexcept
+{
+    if (overlay.valid()) {
+        close_legacy_overlay_handle_ignoring_status(anchor, overlay);
+        overlay = {};
+    }
+}
+
+void close_legacy_overlay_handles_and_reset(
+    Node& anchor,
+    pp::ui::NodeHandle& popup_overlay,
+    pp::ui::NodeHandle& tick_overlay) noexcept
+{
+    close_legacy_overlay_handle_and_reset(anchor, popup_overlay);
+    close_legacy_overlay_handle_and_reset(anchor, tick_overlay);
+}
+
 void close_legacy_overlay_handles_if_open(
     Node& anchor,
     const pp::foundation::Result<pp::ui::NodeHandle>& popup_overlay,

src/legacy_ui_overlay_services.h

@@ -111,6 +111,13 @@ void close_legacy_popup_overlay(Node& node) noexcept;
 void close_legacy_overlay_handle_ignoring_status(
     Node& anchor,
     pp::ui::NodeHandle overlay) noexcept;
+void close_legacy_overlay_handle_and_reset(
+    Node& anchor,
+    pp::ui::NodeHandle& overlay) noexcept;
+void close_legacy_overlay_handles_and_reset(
+    Node& anchor,
+    pp::ui::NodeHandle& popup_overlay,
+    pp::ui::NodeHandle& tick_overlay) noexcept;
 void close_legacy_overlay_handles_if_open(
     Node& anchor,
     const pp::foundation::Result<pp::ui::NodeHandle>& popup_overlay,

src/main.cpp

@@ -932,8 +932,8 @@ int main(int argc, char** argv)
     wglMakeCurrent(NULL, NULL);
 
     running = 1;
-    App::I->render_thread_start();
+    App::I->runtime().render_thread_start(*App::I);
-    App::I->ui_thread_start();
+    App::I->runtime().ui_thread_start(*App::I);
 
 #ifdef _DEBUG
     glad_set_pre_callback(_pre_call_callback);
@@ -1035,8 +1035,8 @@ LRESULT CALLBACK WndProc(HWND hWnd, UINT msg, WPARAM wp, LPARAM lp)
         running = 0;
         if (hmd_renderer.joinable())
             hmd_renderer.join();
-        App::I->ui_thread_stop();
+        App::I->runtime().ui_thread_stop();
-        App::I->render_thread_stop();
+        App::I->runtime().render_thread_stop();
         App::I->terminate();
         delete App::I;
         PostQuitMessage(0);

src/node.cpp

@@ -42,7 +42,7 @@
 void Node::app_redraw()
 {
     App::I->redraw = true;
-    App::I->render_cv.notify_all();
+    App::I->runtime().notify_render_worker();
 }
 
 void Node::watch(std::function<bool(Node*)> observer)

src/node_dialog_cloud.cpp

@@ -68,7 +68,7 @@ void NodeDialogCloud::init_controls()
     btn_cancel = find<NodeButton>("btn-cancel");
     pp::panopainter::bind_legacy_click_destroys_node(*btn_cancel, *this);
     container = find<Node>("files-list");
-    std::thread(&NodeDialogCloud::load_thumbs_thread, this).detach();
+    load_thumbs_worker_ = std::jthread(&NodeDialogCloud::load_thumbs_thread, this);
 }
 
 void NodeDialogCloud::loaded()
@@ -79,6 +79,7 @@ void NodeDialogCloud::removed(Node* parent)
 {
     NodeBorder::removed(parent);
     closed = true;
+    load_thumbs_worker_.request_stop();
 }
 
 NodeText* NodeDialogCloud::create_loading_status_text()
@@ -143,24 +144,30 @@ std::vector<NodeDialogCloudItem*> NodeDialogCloud::create_cloud_file_items(const
     return nodes;
 }
 
-void NodeDialogCloud::load_thumbs_thread()
+void NodeDialogCloud::load_thumbs_thread(std::stop_token stop)
 {
 #if WITH_CURL
     BT_SetTerminate();
-    CURL *curl = curl_easy_init();
+    auto curl = std::unique_ptr<CURL, decltype(&curl_easy_cleanup)>(curl_easy_init(), curl_easy_cleanup);
     std::string res;
     if (curl)
     {
+        if (stop.stop_requested() || closed)
+            return;
+
         auto* text = create_loading_status_text();
         auto* align = text->m_parent;
 
-        if (!load_cloud_file_list(curl, res, *text))
+        if (!load_cloud_file_list(curl.get(), res, *text))
         {
             return;
         }
         
         pp::panopainter::destroy_legacy_node(*align);
 
+        if (stop.stop_requested() || closed)
+            return;
+
         LOG("CLOUD LIST: %s", res.c_str());
 
         auto names = split(res, ',');
@@ -171,13 +178,12 @@ void NodeDialogCloud::load_thumbs_thread()
         {
             const auto& n = names[i];
             auto* node = nodes[i];
-            if (closed)
+            if (stop.stop_requested() || closed)
                 break;
 
-            if (!load_cloud_thumb(curl, n, node, res))
+            if (!load_cloud_thumb(curl.get(), n, node, res))
                 break;
         }
-        curl_easy_cleanup(curl);
     }
 #endif //CURL
 }

src/node_dialog_cloud.h

@@ -6,6 +6,8 @@
 #include "node_text_input.h"
 
 #include <vector>
+#include <stop_token>
+#include <thread>
 
 class NodeDialogCloudItem : public NodeBorder
 {
@@ -46,8 +48,11 @@ public:
     void init_controls();
     virtual void loaded() override;
     virtual void removed(Node* parent) override;
-    void load_thumbs_thread();
+    void load_thumbs_thread(std::stop_token stop);
     NodeText* create_loading_status_text();
     bool load_cloud_file_list(CURL* curl, std::string& response, NodeText& status_text);
     std::vector<NodeDialogCloudItem*> create_cloud_file_items(const std::vector<std::string>& names);
+
+private:
+    std::jthread load_thumbs_worker_;
 };

src/node_panel_stroke.cpp

@@ -722,14 +722,10 @@ void NodePanelStroke::execute_stroke_control_plan(const pp::app::BrushStrokeCont
 
 void NodePanelStroke::close_popup_overlay_handles() noexcept
 {
-    if (m_popup_overlay_handle.valid()) {
+    pp::panopainter::close_legacy_overlay_handles_and_reset(
-        pp::panopainter::close_legacy_overlay_handle_ignoring_status(*this, m_popup_overlay_handle);
+        *this,
-        m_popup_overlay_handle = {};
+        m_popup_overlay_handle,
-    }
+        m_tick_overlay_handle);
-    if (m_tick_overlay_handle.valid()) {
-        pp::panopainter::close_legacy_overlay_handle_ignoring_status(*this, m_tick_overlay_handle);
-        m_tick_overlay_handle = {};
-    }
 }
 
 kEventResult NodePanelStroke::handle_event(Event* e)

src/node_popup_menu.cpp

@@ -4,6 +4,7 @@
 #include "node_popup_menu.h"
 #include "node_button_custom.h"
 #include "app.h"
+#include <memory>
 
 Node* NodePopupMenu::clone_instantiate() const
 {
@@ -33,6 +34,8 @@ kEventResult NodePopupMenu::handle_event(Event* e)
     case kEventType::MouseDownL:
         break;
     case kEventType::MouseUpL:
+        {
+            auto self = std::static_pointer_cast<NodePopupMenu>(shared_from_this());
             if (m_mouse_inside)
             {
                 for (int i = 0; i < m_children.size(); i++)
@@ -40,12 +43,13 @@ kEventResult NodePopupMenu::handle_event(Event* e)
                     if (m_children[i]->m_mouse_inside)
                     {
                         if (on_select)
-                        on_select(this, i);
+                            on_select(self.get(), i);
                         break;
                     }
                 }
             }
             close_popup();
+        }
         break;
     default:
         return kEventResult::Available;
@@ -61,13 +65,19 @@ void NodePopupMenu::added(Node* parent)
     m_mouse_ignore = false;
     m_flood_events = true;
     m_capture_children = false;
+    auto self = std::static_pointer_cast<NodePopupMenu>(shared_from_this());
     for (int i = 0; i < m_children.size(); i++)
     {
         if (auto b = std::dynamic_pointer_cast<NodeButtonCustom>(m_children[i]))
         {
-            b->on_click = [this, i](Node* target) {
+            std::weak_ptr<NodePopupMenu> weak_self = self;
-                if (on_select)
+            b->on_click = [weak_self, i](Node* target) {
-                    on_select(this, i);
+                auto self = weak_self.lock();
+                if (!self) {
+                    return;
+                }
+                if (self->on_select)
+                    self->on_select(self.get(), i);
             };
         }
     }

test.cpp

@@ -0,0 +1,7 @@
+#include <memory>
+struct A{};
+int main(){
+    std::shared_ptr<A> p;
+    A* a = nullptr;
+    p = a;
+}