A change of direction#

Sat Mar 14 18:20:42 2026

I've pivoted twice on Eculocate since the previous blog post: first a minor change of direction and then followed by a much bigger one.

I'll start with the bigger one: I've stopped working on it. This is because I am 95% sure it won't actually work with my motorbike. What's my basis for this?

• consider: "gear indicators" for motorbikes with k-line diagnostics plug into the ECU diagnotsic port (in fact, the honda k-line protocol was pretty much reverse-engineered using one of them). The gear indicators for bikes of the same age as mine tee into the speed sensor connector and also tap the rpm input to the ECU. Why would they do this if they could just use the diagnostic port?

• if you have a fancy workshop diagnostic tool, the appropriate cable to use it with a 2001 CBR600F is described as a blink code cable - which rather suggests there isn't any more data provided by the ECU than ... blink codes

I don't know definitively that it won't work. I do know that if it doesn't work and it releases any kind of magic smoke, replacing the ECU is going to cost a pretty huge amount of money. So, Im just not going to.

That's the major pivot.

The minor pivot, which preceded it, was when I realised that I can't use the Python Bleak library on my phone with Termux, because its Android support uses a different Python-for-Android backend than the Termux linux-userland-for-android. Argh. So, I decided to stop trying to do Android, and instead dusted off Biscuit which is my previously abandoned project to make a bike computer using a Motorola G4 Play phone (it has a removable battery and is old enough to be reasonably well supported on mainlin Linux) running Mobile Nixos. So I added an eculocate module and ported the eculocate gui client from termux:gui to Gtk and then I thought I'd better check a few things and then I had the major pivot. So, that's the chronology.

What now?

• I've spent a few days bringing the stalled Mobile Nixos device back up to date: current nixpkgs, rebase against current mobile-nixos version, upgrade to later kernel, etc. Will have to figure out what else is outstanding before I can reasonably ask for a review/merge.

• and some time on the "maps" app in biscuit, mostly just to make the scrolling less lurchy as the gps moves around

• and some time on the quite unrelated task of bringing Liminix back up to date so it can be built using a current Nixpkgs version rather than the last release but one.

The plan currently is to get Biscuit into some kind of shape that I can dogfood it:

• at minimum, show a gpx route overlaid on the map

• extra credit, add kind of turn-by-turn navigation/rerouting if the rider should stray from the gpx path

• given that (last time I tried it) the phone wouldn't acquire a GPS lock in less than ten minutes, there may still be a role for reusing (parts of) eculocate to provide external GPS and maybe some other sensors like IMU - perhaps even buy one of those tap-the-speed-sensor gear indicators for its wiring and expose the bike engine speed. Project GXXR has lots of interesting stuff on this subject.