Furnace previous_recipe ids are nested under prev.name.name in this context. Restore the original access path and leave an inline maintainer note to prevent repeat regressions.
Dual filter API methods (setInstanceSurfaceFilters and setInstanceSurfaceFilter) both exist for backward compatibility with different plugin implementations. Added comment explaining why both are checked in fallback pattern.
- Add warning logging in renderChunkToPixels catch block before fallback
- Switch maxTick filtering from skip/continue to early break (append-only optimization)
- Use existsSync for directory existence check in loadExistingTagContent
- Use player.physical_surface/physical_position for accurate tracking when in vehicles/remotes
- Emit session start record immediately on on_player_joined_game
- Make storage.minimap initialization consistent (init() ensures all fields exist)
Task 3 inadvertently changed chart tag data schema from start_tick/end_tick
interval model to single tick field. This restores the original interval
semantics for chart tags while preserving legitimate Task 3 fixes.
Reverted:
- create_chart_tag_data(tick) -> create_tag_data(start_tick, end_tick)
- Single-emission in on_chart_tag_modified -> dual-emission (end old, start new)
- All chart tag event handlers use interval model again
Preserved from Task 3:
- prev.name.name fix (both locations)
- Hot-loop local aliases (math_floor, bit32_rshift, table_insert)
- IIFE removal for recipe icon lookup
- Recipe data single tick field (within Task 3 scope)
- Fix prev.name.name bug: previous_recipe.name is RecipeID, not nested
- Add hot-loop local references (math_floor, bit32_rshift, table_insert)
- Remove IIFE pattern for icon lookup, use inline logic
- Rename create_tag_data to create_chart_tag_data for consistency
- Use recipe nilness + single tick field instead of start_tick/end_tick split
From cursor review: Insufficient bounds checking
for pixel change records (type 2) in renderTileToPixels,
extractAvailableTicks, and parseTileData. The code
checks for 6 bytes of header data but reads 10 bytes
(tick, pixelCount, newColor, oldColor), potentially
leading to reading out of bounds. The check should
be offset + 10 > tileData.length.
Danielv123