- led-simulator: the rig runs the C++ led-driver as WebAssembly (sim-driver/),
one single-threaded worker per bar, no cross-origin isolation needed. Rooms
(one JSON per room) with an on-page editor, and a pattern gallery. The
Python server, Pyodide boot and ring buffer are removed.
- sim-controller sim_rig.c: serves /sim/ from an embedded pack, the page's
WebSocket and the rooms/flash API. The desktop build embeds it too.
- bridge.c: a "browser" bridge kind that carries ESP-NOW frames to the page;
devices announced through it are marked sim.
- UI: Simulator in the Menu, split view with resize and pop-out, swipe
between controller and rig on phones; SIM/REAL badges in Devices.
- Fades: preset pushes and Wi-Fi driver bodies always carry fade and
beat_time, so "off" no longer inherits an earlier fade.
- Also the sequences-v2 work in progress: desktop port, bridge firmware,
serial bridge, Wi-Fi drivers, sequence timeline, audio and API changes.
- led-driver: ESP-NOW link, beat sync, control messages, sim hooks.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
- Sequences: advance_mode "time" makes each step last its `seconds`
(0.1–3600) instead of beats; editor gets a "Steps last" selector.
- UI: search box in the groups modal; searchable add pickers for zone
groups and presets.
- Preset "try" sends only to the current zone's groups, never broadcast.
- led-driver: max_brightness ceiling from device config; follow a
controller that restarted its beat counter.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Profiles, presets and sequences export and import as main's bundles.
Zones carry group_ids, preset_group_ids, sequence_ids, content_kind and
brightness; profiles seed rainbow as colour_cycle, and deleting a
profile also deletes its sequences. Groups, zones and sequences written
by older versions are migrated at boot.
/api/audio (status, event stream, start/stop, reset, anchor-bar,
devices) maps main's audio detector onto the mic's beat detector. Audio
start/stop picks whether mic beats or the BPM clock time sequences; the
clock's beats also go to the drivers while the mic is off.
/ws now greets from a post-handshake callback (httpd does not call the
handler for the upgrade).
The web UI is main's, plus v2's outputs editor, list search and
hostname/static IP settings; the bridge, Wi-Fi AP and led-tool parts
are gone. API tests follow main's behaviour (119 pass on the board).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Groups (registry of driver MACs, optionally scoped to a profile) become
the drivers' group membership, alongside z<zone id> for zones that still
name drivers. Zones play to their group_ids. Group and device routes for
driver config, brightness (global x groups x device x zone) and identify.
The push path takes main's message shape: select as a list, groups as a
filter (one GROUP_CMD per group, or JSON when a preset has a background),
short keys so JSON fits a frame.
Sequences follow main: lanes of {preset_id, beats} steps played to each
lane's groups, advanced by the mic's beats or a clock at
audio_simulated_bpm; a switch waits for the next beat or downbeat; sync-
phase, stop (clears the zone's drivers) and status. This replaces the
time-based per-zone sequencer and scene playback, as in main.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>