Led astray#

Sat Jan 30 12:46:06 2021

Topics: nix

There are probably better ways to satisfy the need for a wall light than

cutting up a 150 LED Neopixel strip, sticking the pieces to a piece of aluminum, then soldering wires from each row to the next
building a ESP8226 circuit on stripboard to drive it
sending 10x15 arrays of RGB over MQTT to specify colours

The led strip came from my Christmas lights, as did the circuit - now transferred from breadboard to a more permanent soldered stripboard. The Arduino sketch, known as Dolores ("do lo-res") is all new, though, and parts of it (the bits that are plain C++ not Arduino code) are even unit-tested.

Perhaps worth noting

10x15 pixels with a byte each for R G B is 450 bytes of data, which is more than the default max packet size of the Arduino MQTT library I'm using. Fix this with a call to client.setBufferSize with a more appropriate value
after the first time I'd put it all together and installed it on the wall, I realised that software updates were going to be considerable faff, given they require unplugging the Wemos D1 from its circuit and attaching it to USB. OTA updates to the rescue! This makes the device run a HTTP server which new images can be POSTed to. There are better ways to do this - I'd rather it were an HTTP client which fetched images from a known URL, instead of accepting anything from anyone that can reach its port 80 - but that can wait until I have some kind of CI setup.
the Nix environment I'm using for Arduino IDE only works with nix-shell not nix-build. I'd like it to behave as a regular derivation so I can have a CI setup

But now I can send it any picture I like. Here's a digital clock -

while sleep 2 ; do \
  ( convert -support .5 -depth 8 -fill red \
            -font DejaVu-Sans-Condensed -pointsize 7 -size 15x10! \
	    -draw "text 0,7 '$(date +%H%M)'" -sharpen 8 'xc:#333'  RGB:- | \
    tee out.rgb | \
    nix run nixos.mosquitto -c mosquitto_pub  -h mymqttserver.lan \
    	    		         -s --username username --pw password  \
				 -t effects/84F3EBE5E57C/image ) ; \
done

Her name is Oreo#

Fri Feb 12 12:47:14 2021

Topics: android

I am at a point somewhere along the way of a long yak-shaving journey ("I wonder if the MSM8953 SoC supports ANT+ in hardare and it just needs enabling in the Android software?") which involves rebuilding Android for my phone. There seem to be very few ROMs for the Moto G5 Plus ("potter") that are still being maintained, and even fewer of them that have build instructions that (a) work as described; (b) comprise fewer than 30 manual steps. Do Android ROM developers not do CI builds? If this was all GPL I could make a strong case it's in breach. ("The source code for a work means the preferred form of the work for making modifications to it [...] all the source code for all modules it contains, plus any associated interface definition files, plus the scripts used to control compilation and installation of the executable."). Yes, I know, volunteers, spare time, teeth of gift horses, etc etc. End rant

Anyway, so far I am doing the following and it has not yet blown up

using the docker image from LineageOS for Microg
invoking it as follows

docker run \
    -e "BRANCH_NAME=lineage-15.1" \
    -e "DEVICE_LIST=potter" \
    -e "SIGN_BUILDS=false" \
    -e "SIGNATURE_SPOOFING=restricted" \
    -e "INCLUDE_PROPRIETARY=false" \
    -e "WITH_SU=true" \
    -e "CUSTOM_PACKAGES=GmsCore GsfProxy FakeStore MozillaNlpBackend NominatimNlpBackend com.google.android.maps.jar FDroid FDroidPrivilegedExtension " \
    -v "/$HOME/src/android/l4mg/lineage:/srv/src" \
    -v "/$HOME/src/android/l4mg/zips:/srv/zips" \
    -v "/$HOME/src/android/l4mg/logs:/srv/logs" \
    -v "/$HOME/src/android/l4mg/cache:/srv/ccache" \
    -v "/$HOME/src/android/l4mg/keys:/srv/keys" \
    -v "/$HOME/src/android/l4mg/manifests:/srv/local_manifests" \
    lineageos4microg/docker-lineage-cicd

with this manifest in $HOME/src/android/l4mg/manifests/roomservice.xml

<?xml version="1.0" encoding="UTF-8"?>
<manifest>
	<project name="LineageOS/android_external_bson" path="external/bson" remote="github" revision="cm-14.1" />
	<project name="LineageOS/android_device_qcom_common" path="device/qcom/common" remote="github" revision="lineage-15.1" />
	<project name="LineageOS/android_packages_resources_devicesettings" path="packages/resources/devicesettings" remote="github" revision="staging/lineage-15.1" />
	<project name="boulzordev/android_device_motorola_potter" path="device/motorola/potter" remote="github" revision="lineage-15.1-64" />
	<project name="boulzordev/android_vendor_motorola_potter" path="vendor/motorola/potter" remote="github" revision="lineage-15.1-64" />
	<project name="boulzordev/android_kernel_motorola_msm8953" path="kernel/motorola/msm8953" remote="github" revision="lineage-15.1-64" />
</manifest>

Worthy of note:

I found the manifest on the Interwebz by use of a search engine. I had to edit the revision attribute of android_device_qcom_common from staging/lineage-15.1 to lineage-15.1
because I have a (presumed working) example manifest for 15.1 I am not attempting to build any newer version. One change at a time, hey
in my limited understanding, the manifest tells the build system where to find Moto proprietary files, and the INCLUDE_PROPRIETARY=false build argument tells the L4microg build scripting not to try and download them from "TheMuppets". This is important because the muppets don't have them for my device.
getting F-droid and Microg baked in is probably no bad thing

As I write this, it has been building for ~ 30 minutes and not fallen over yet, so things are looking vaguely promising

[  8% 8343/95970] target thumb C++: libv8src_32 <= external/v8/src/builtins/builtins-constructor.cc

Apparently it's done 8% of something

Update: Fri Feb 12 19:57:48 GMT 2021

It works! It built a buch of stuff in zips/potter/ including the file lineage-15.1-20210212-UNOFFICIAL-potter.zip which I installed from Recovery using the instructions at https://wiki.lineageos.org/devices/peregrine/install Yes, that's not the same phone, but it's close enough to use the same procedure

⟪ Jan 2021 Mar 2021 ⟫