Sparky's Plans

Sparky's Plans

Public to-do lists and goals for current projects.

TRMNL E-Ink Dashboard

Deprecated. The space runs a different e-ink display now. These items are not being worked on.

TODO

  • [ ] Integrate occupancy display showing who is currently in the space
    • Option 1: Pull badge-in data from ButterflyMX
    • Option 2: Pull presence data from Home Assistant
    • Determine best data source and create custom plugin

Transit Tracker

Done

  • [X] Set up DNS for the Raspberry Pi API endpoint → transittracker-pi.int.dma.computer (10.20.100.80)
  • [X] Set up DNS for the display device → transittracker-display.user.dma.computer (10.20.110.197)
  • [X] Set up static IP for the display device (10.20.110.197)
  • [X] Carousel: cycle through trips one (or more) at a time
  • [X] Walking-time offset per stop (already in upstream firmware as time_offset)
  • [X] “Trips remaining today” indicator: contributed a new tripsRemainingToday field to the API for GTFS-static feeds, plus firmware support that prefers it over the OBA fallback. Display only shows the indicator when service is winding down (remaining_trips_threshold: 5).
  • [X] Bring up local st_gtfs feed (Sound Transit GTFS-static + GTFS-RT) so the display gets truthful end-of-day counts while keeping real-time predictions
  • [X] Display fixes during testing: 180° rotation, fill the 32px height with 3 trips, prevent (-N) indicator from overlapping the realtime icon, abbreviate long headsigns
  • [X] Upstream contribution: PR #65 to quote partition table names so hyphenated feed codes (like st-gtfs) don't break the sync
  • [X] Stream Deck preset buttons: a 15-key Stream Deck on the Pi runs streamdeck-transit in Docker. Keys call select.select_option on select.transit_tracker_cycling_transit_preset, and the deck polls Home Assistant back so the live preset stays lit even when someone switches it from the web UI.
  • [X] Presets in place: Link (1 Line + 2 Line combined), Bus 8, Bus 10, Bus 12, First Hill Streetcar
  • [X] Walking-time offsets calculated from the sign's GPS to each stop in the active preset (sign_location + walk_speed_ms), so offsets follow the preset automatically
  • [X] Display power and brightness exposed to Home Assistant (switch.hangout_transit_tracker_cycling_display_power, number.hangout_transit_tracker_cycling_display_brightness), so a deck key can blank the sign without cutting power
  • [X] Fixed a King County Metro outage: a sync_lock row orphaned since July blocked that feed from syncing for 60 days, its schedule data expired, and every KCM route went blank. Forced a resync and restarted the API to clear cached empty results.

TODO

  • [ ] Ship the API fixes for that outage: reclaim sync locks older than six hours, and retire a feed's cached data after an import so a resync is visible without a restart (branch fix/stale-sync-lock)
  • [ ] Flip the walking pip sign from +7w to -7w; - matches the actual math (arrival minus walk equals leave-by)
  • [ ] Use the spare matrix row as a header showing the active preset name, so it is obvious which preset is on screen
  • [ ] Return the Orange Pi lent for the button project. The Stream Deck runs on the existing transit Pi, so it was never needed.
  • [ ] Label the Stream Deck keys physically, or add a legend nearby, so the presets are usable without asking anyone

Big LED Signs (transit display embiggening)

Two salvaged LED signs are earmarked for a larger version of the arrivals board. Asks and offers collected in #facilities > “next train display embiggening idea”.

Done

  • [X] Identified sign 1: iNOVA C08160S2, two lines of 160×8, a call-center wallboard. Recovered 12 original PDFs from the Wayback Machine; instref.pdf is the useful one.
  • [X] Confirmed the RJ11 “net in” jack is an RS485 daisy-chain port, 9600 baud by default, not a phone line
  • [X] Identified sign 2: Adaptive ALPHA 9200TRI, tri-color. Its protocol is public (Adaptive doc 9708-8061) and the python alphasign library already speaks it. The ethernet jack is a Lantronix serial bridge, so it should be drivable once on the network.
  • [X] Established a fallback for the iNOVA: its native command set is not published anywhere, so the panels can be driven directly instead, per the eyadid/esp-led-controller project
  • [X] A USB to RS485 adapter was donated and is at the space, on top of the yellow rack

TODO

  • [ ] Collect the donated RS485 adapter from the yellow rack
  • [ ] Find a sacrificial RJ11 cable and try talking to the iNOVA sign at 9600 baud
  • [ ] Get the ALPHA 9200TRI into the space. It currently fails self-test and its fan is very loud, though all the LED panels light. Then try alphasign over the Lantronix bridge (data on TCP 3001, config via telnet on 9999).
  • [ ] Decide where it mounts. Suggestions so far: above the cork board, or roughly 8ft up on the north wall above the toolbox.
  • [ ] Pick which sign to target first. The ALPHA is far less work because its protocol is documented.

Layout requests from the thread

  • [ ] Show four trains at once, which the current 32px display cannot do
  • [ ] Put a clock on it. The existing space clock is unpopular. One suggestion was trains on the left, clock on the right.
  • [ ] Colocate with the 1+2 line PCB so all the transit displays live in one area
  • [ ] Scrolling messages on the two-line sign rather than static text
  • [ ] A “days since someone tried to drive on the track” counter

WhereTrain (ESP32 Transit Tracker)

TODO

  • [ ] Integrate with the Raspberry Pi transit-tracker API (transittracker-pi.int.dma.computer) instead of calling OneBusAway directly
    • Currently the ESP32 polls OneBusAway API directly from main.c
    • Goal: Point it at the existing transit-tracker API so all transit data flows through one source
    • Reduces redundant API calls and centralizes API key management on the Pi