Fix connection-status, drawing-tool and node-tracking defects - #176
Merged
amorfinv merged 7 commits intoAug 13, 2026
Merged
Conversation
The network client auto-selects only the very first node it ever sees, and removal-failover needs a surviving node to switch to. When the last node vanishes (e.g. its process crashes) act_id keeps pointing at the dead node, so a node added afterwards was never activated: the actonly subscriptions (ACDATA/ROUTEDATA) stayed bound to the dead node, no traffic flowed, and the data-flow timeout tore down an otherwise live connection. _on_node_added now re-activates when the client's active node is gone, and the local `import traceback` calls are hoisted to a module-level import. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
get_current_data() built its own node_info payload straight from the internal stores, which keep binary node/server IDs. Socket.IO ships those as binary attachments and the frontend NodeData type can't read them, so a client that joined between node_info events saw unusable node data. The serialization used by the node_info event is extracted into NodeManager.serialize_node_info() and both paths now go through it. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
…disconnect Three related connection-status defects: - A deliberate disconnect (QUIT, server_disconnected, Stop in the integrated build) left the armed no-data timer running, so it fired up to 5s later and echoed a spurious "No data received - connection may be lost" warning. setBlueSkyConnected(false) now cancels the timer and clears receivingData, which also fixes receivingData sticking true forever after a WebSocket drop. - Data events already in flight when the disconnect landed flipped the status straight back to connected, re-arming that timer and flashing a phantom "Connected". expectDisconnect() opens a short grace window in which data is ignored; SocketManager also drops in-flight siminfo/acdata frames so they can't repaint traffic the disconnect just cleared. Data still arriving after the window (i.e. the disconnect request failed) reconnects as usual. - SettingsModal never propagated the configured host, so the header status string reported its 'localhost' default forever. Also drops the unused ConnectionState duplication (webSocketState / blueSkyState mirrored the booleans) and dead accessors. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Starting shape drawing while a route draw was in progress (or either while placing an aircraft) left both tools listening, so one map click was consumed twice: waypoints landed in a route the user thought they had left, and the shared drawing banner showed whichever tool started last. A small drawingExclusion module holds the single active claim; claiming cancels the previous owner's draw. Route and shape drawing claim the map before showing their banner so the outgoing tool's teardown can't hide it. Alongside: - AircraftInteractionManager's route-only predicate becomes a general setDrawingToolActiveCheck, so empty-map clicks during shape drawing and aircraft placement no longer deselect the selected aircraft. - Teardown no longer bails out when the map is already gone, which used to strand the document-level keydown/Escape listener and the navaid highlight. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
BlueSky can deliver NaN or out-of-range lat/lon (e.g. after a MOVE past lat 90). The scene origin was seeded from — and recentred on — every record, so one bad aircraft poisoned the origin shared by all mesh transforms and the mercator camera matrix, and MercatorCoordinate.fromLngLat then threw on every data tick and every frame. updateSceneOrigin now filters through isValidCoordinate before seeding or recentring, mirroring the per-aircraft guard the renderers already apply. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
A model request still in flight when clearAll() ran resurrected the caches it had just dropped: the model was cached and onModelLoaded fired against a torn down renderer, leaking the GLB's geometry and materials. The load callback now checks that its entry is still registered and disposes the result otherwise. clearCache() also dropped its recorded raw dimensions, which previously survived the models they described. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
Aircraft3DFleet repeated the globe/mercator transform branch in four places; it now goes through one applyTransform helper. MapControlsPanel's three near-identical button handlers collapse into a withManager helper, and the zoom/reset paths lose their duplicated null checks and narration-only logs. No behaviour change. Co-Authored-By: Claude Opus 5 (1M context) <noreply@anthropic.com>
amorfinv
deleted the
fix/connection-status-drawing-tools-and-node-tracking
branch
August 13, 2026 20:27
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
1 participant
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.
Seven commits, each a self-contained fix (plus one no-behaviour-change cleanup at the end).
Backend (proxy)
act_idkept pointing at the dead node, the actonly subscriptions (ACDATA/ROUTEDATA) never moved, and the data-flow timeout tore down a live connection.node_infoin theinitial_datasnapshot. It was built straight from the internal stores, which hold binary node/server IDs — unusable by the frontendNodeDatatype. Both paths now shareNodeManager.serialize_node_info().Frontend
receivingDatastale after a WebSocket drop; in-flight data events flipped the status straight back to connected. Adds a shortexpectDisconnect()grace window, and drops theConnectionStatefields that merely mirrored the booleans.SettingsModalnow propagates the configured host so the header stops reportinglocalhost.drawingExclusionmodule holds the single active claim. Empty-map clicks during shape/aircraft drawing no longer deselect the aircraft, and teardown with the map already gone no longer strands document-level listeners.MOVEpast lat 90) poisoned the shared scene origin, makingMercatorCoordinate.fromLngLatthrow on every tick and frame.clearAll()had just dropped and leaked geometry/materials.New tests cover each fix.
🤖 Generated with Claude Code