Server-side optimizations:
- Remove all Process.send_after delays - use immediate send()
- High zoom (10+): Load everything in single batch (up to 500 packets)
- Low zoom: Load all batches immediately without delays
- Remove debug logging for performance
Client-side optimizations:
- Remove chunking delays and requestAnimationFrame
- Process all packets immediately without setTimeout
- Remove 10-packet chunk limit - process everything at once
Result: Historical packets now load as fast as possible
without artificial delays or chunking bottlenecks.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Key fixes:
- Combine bounds_changed and update_map_state into single event
- Remove duplicate sendBoundsToServer() calls
- Add programmaticMove flag to prevent event loops
- Only process bounds updates when bounds actually change
- Debounce events with 300ms timeout
This eliminates:
- Duplicate server events for same user action
- Event loops when server updates URL
- Unnecessary bounds processing
- Excessive server load
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Features implemented:
- Parse URL parameters on mount (lat, lng, z)
- Update URL when map moves or zooms
- Restore map state from URL parameters
- Handle URL changes via handle_params
- Validate and clamp coordinate values
URL format: /?lat=40.7128&lng=-74.0060&z=12
Benefits:
- Shareable map links with exact position
- Browser back/forward navigation support
- Bookmarkable map locations
- Deep linking to specific map views
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Critical fixes:
- Replace old load_historical_packets_for_bounds with progressive loading
- Fix handle_info_initialize_replay to use progressive loading
- Fix process_bounds_update to use progressive loading
- Add debug logging to track batch loading
High zoom optimizations:
- Very zoomed in (15+): 10 packets per batch, 10ms delays
- Moderately zoomed (12-14): 20 packets per batch, 20ms delays
- Medium zoom (8-11): 35 packets per batch, 35ms delays
- Use linear delays instead of exponential for faster loading
This ensures progressive loading is actually used and optimized
for high zoom levels where viewport is small.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
Key improvements:
- Fix spatial query order: Apply bounds filtering BEFORE limit/offset
- Add zoom-level based batch sizing (20-100 packets per batch)
- Dynamic batch count based on zoom level (2-5 batches)
- Faster loading delays for high zoom levels (25ms vs 50ms)
- Include zoom level in cache keys for better cache efficiency
Performance benefits:
- Zoomed in far (zoom 15+): 2 batches of 20 packets each
- Moderately zoomed (zoom 12-14): 3 batches of 35 packets each
- Medium zoom (zoom 8-11): 4 batches of 50 packets each
- Zoomed out (zoom <8): 5 batches of 75-100 packets each
This ensures packets are loaded from the current viewport first,
dramatically reducing load times when zoomed in.
🤖 Generated with [Claude Code](https://claude.ai/code)
Co-Authored-By: Claude <noreply@anthropic.com>
